Search Results

Yesterday's news cycle was quiet on hype but loud on execution. Big Tech dropped earnings, governments moved, and agentic capabilities kept advancing. Here's what actually happened and why it matters for anyone trying to stay in the driver's seat at work. Zuckerberg directly ties AI costs and efficiency gains to Meta's upcoming layoffs In an internal all-hands Q&A, Mark Zuckerberg told employees that rising compute expenses and workflow improvements from AI were key factors behind the planned reduction of roughly 8,000 roles. This wasn't vague corporate speak — it was an explicit link between the massive capex Meta is pouring into AI infrastructure and the resulting headcount adjustments. The company has been scaling AI across WhatsApp, Messenger, and internal tools, with business AI conversations now hitting 10 million per week. That kind of efficiency doesn't come free, and it's already reshaping org charts at one of the largest tech employers. Musk wraps testimony in the OpenAI federal trial Elon Musk finished his testimony in the ongoing lawsuit over OpenAI's shift from nonprofit roots to a capped-profit structure. The case centers on mission alignment, governance, and whether the original charter still holds as the company scales commercially. Court records and public statements from both sides are on file — no speculation needed. When a vendor changes its incentives or ownership model, reliability and alignment can shift. When your AI provider updates terms or shifts strategy, ask one question: "Does this keep the model working for my team's goals, or theirs?" If the answer isn't immediately obvious, it's time to test alternatives and add a second vendor. WRITER ships event-based triggers for enterprise AI agents The platform rolled out native support for autonomous, event-driven workflows with built-in governance controls. Instead of waiting for a prompt, agents can now react to triggers — new lead in Salesforce, status update in Slack, or a completed approval — and take defined next steps while keeping humans in the loop for review or escalation. This moves agents from chat toys to production tools. The difference between leverage and liability is deliberate design. Pick one repeatable workflow, map it end-to-end, set one clear trigger, define success as "human time saved with zero quality drop," and document what the agent handled versus what needed your judgment. That template becomes the pattern your whole team reuses. Anthropic and OpenAI endorse the Warner-Budd Workforce Transparency Act Both companies publicly backed proposed legislation that would require organizations to disclose when AI plays a material role in hiring, performance reviews, or other workforce decisions. The bill is still early in the legislative process, but two frontier labs on record supporting it signals that transparency around AI-influenced people decisions is moving from nice-to-have to expected. U.S. Q1 GDP growth beat expectations partly on AI-related business investment Commerce Department data showed stronger-than-expected 2.0% growth, with AI-driven equipment purchases and intellectual-property investment providing a measurable lift. Trade and inventories offset some of the gain, but the AI spending component was real and verified. Macro validation matters, but those gains only show up in your P&L when humans direct the spend with clear metrics. Treat every new AI tool like any other vendor system — require basic security and data handling details, and set a simple "what if it hallucinates" response plan. White House convenes tech firms on AI-driven cybersecurity threats Senior officials met with leading companies to discuss both the offensive and defensive uses of powerful new models. Details remain under NDA, but the focus was collaboration on emerging risks rather than immediate new mandates. Security is now an agentic battlefield. Every AI tool you adopt needs the same scrutiny you'd give any external system. Add one line to your team's usage guideline: "Verify output against primary sources before it touches a customer, contract, or production code." Uber burns through its entire 2026 AI budget in just four months Uber's CTO confirmed the company exhausted its full-year AI allocation by April. Roughly 5,000 engineers received access to tools like Anthropic's Claude Code and Cursor, with adoption hitting 95% monthly usage and 70% of committed code now AI-generated. Per-engineer API costs ran $500–$2,000 per month. The productivity gains were real — the budget assumptions were not. This is the clearest real-world example yet of what happens when AI tools work too well without a cost framework to match. The lesson is immediate: AI spend is no longer optional, but uncontrolled AI spend is an unnecessary risk. Run a simple audit on your team — actual monthly AI spend per person, which workflows are generating the most value, where you need hard caps or approval gates before you scale further. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

The last 24 hours delivered verified data points from research firms, funding rounds, earnings, and government actions. The picture is consistent: agentic AI is leaving the lab and entering operations at speed. The gap isn't between organizations that believe in AI and those that don't — it's between those building deliberate deployment frameworks and those hoping the tools figure it out. Ninety-two percent of executives expect fundamental change — eighty percent are still supervising every step Genpact and HFS Research surveyed 545 leaders: 92% expect AI to fundamentally shift operations. Spending is projected up 38%. Yet 80% keep agents in supervised mode and only 22% trust agents with broad autonomy. The supervision instinct isn't wrong — it's the only responsible approach at this stage. The problem is when it stays the default indefinitely because no one defined what "enough oversight" looks like for a given task. Supervised agents cost nearly as much as human-only workflows and deliver less throughput. The teams moving fastest have defined explicit handoff criteria: here's what the agent handles autonomously, here's the trigger for human review, here's who owns the escalation. Netomi raised $110M to deploy customer-service agents in production The funding backs agents designed for medium-complexity queries — airline rebooking, insurance claims, order status with exceptions. Accenture is training hundreds of its people on deployment and integration. This isn't beta software — it's production deployments at scale in regulated industries. The useful benchmark for any team evaluating customer-facing AI: track resolution time and satisfaction side by side, not one or the other. Resolution time is easy to optimize and meaningless if customers are unhappy with the outcome. You want both moving in the right direction. Blackstone created BXN1 to concentrate its AI bets Schwarzman's team spun up a dedicated investment structure combining conviction in AI technology with conviction in the infrastructure it runs on — data centers and power generation specifically. A dedicated vehicle suggests a long-time-horizon bet. When institutional capital this size creates a separate structure for a category, it usually means the partners believe it's large and long enough to justify its own governance. The tools you evaluate next year will be better and cheaper because this investment is happening now. The White House blocked Anthropic's Mythos expansion over cyber risk Mythos is Anthropic's frontier research model — capable of autonomously hunting software vulnerabilities and executing attacks. The administration put a speed bump on broader access while security review continues. This is the most important story in the batch for anyone building internal AI governance frameworks. The most capable AI models can be offensive weapons. The question for your team isn't whether you'll use models this powerful — it's whether your access controls, audit logging, and human override mechanisms are designed for what these tools can actually do today. Meta's business AI crossed 10 million conversations per week January: 1 million. Now: 10 million. The growth is happening in WhatsApp business channels, Messenger, and enterprise API integrations. The 10x growth in a single quarter signals the tooling is stable enough for repeated use at scale. For teams that touch external communications: the question isn't whether to test conversational AI in customer channels. It's how to calibrate brand voice, escalation rules, and quality review so the human judgment that makes your customer relationships valuable doesn't get averaged out by the model's tendency toward generic helpfulness. Mercor is paying white-collar workers to document themselves out of a job The platform hires experts at hourly rates to walk through their work in detail — documenting routines, decision frameworks, edge cases — so AI agents can be trained to replicate them. For organizations sitting on significant institutional knowledge — experienced underwriters, senior analysts, seasoned account managers — this is worth taking seriously as a capability transfer mechanism. The value isn't just in the agent it eventually trains. It's in making tacit knowledge explicit and auditable, which is useful whether or not you ever automate the work. Agents are generating a third of internet traffic — and a growing share of attacks Thales security data shows AI-driven bot attacks up 12.5x. Agents now represent a distinct traffic category alongside humans and traditional bots. Authentication, least-privilege access, and audit logs for every agent deployment aren't optional anymore. The same governance framework that protects you from external agent attacks applies to how you deploy your own. If you can't answer "what did this agent do and why" for any action it took, you don't have oversight — you have hope. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

On April 30, Senators Mark Warner and Ted Budd introduced the Workforce Transparency Act — a voluntary framework requiring AI providers and large enterprise customers to submit aggregated, de-identified data on AI usage in workplace tasks to the Department of Labor. Anthropic and OpenAI backed it within 24 hours. That last sentence should give you pause. Who's driving this and why it matters Legislation that moves fast, attracts frontier AI company endorsements, and comes packaged as "voluntary" transparency usually has a shorter shelf life as policy and a longer one as press coverage. When two companies with significant regulatory exposure race to endorse a bill the day after it's introduced, it's worth asking what they're getting out of it — not just what the bill says it does. The Warner-Budd Act asks companies to report how AI is being used in workplace tasks. The data goes to the DOL, gets de-identified and aggregated, and gets published publicly. No mandates. No enforcement. No penalties for opting out. That's not a transparency framework. That's a participation trophy. The structural problem Voluntary disclosure with no enforcement creates a predictable outcome: the organizations with the most significant AI deployments — and the most to reveal — simply don't participate. The data the DOL collects ends up representing the cautious middle of the market, not the edge where the real workforce impact is happening. The result is a report that looks authoritative and tells you very little. The bill's sponsors get credit for addressing AI and jobs. The companies that sign on get to point to federal cooperation. Researchers get a dataset with a selection bias problem baked in from day one. The workers the bill is ostensibly designed to protect get a published report and nothing else. The right framework isn't this one If there's a serious case to be made for AI workforce transparency — and there may be — it requires real enforcement mechanisms, clear definitions of what constitutes material AI involvement in employment decisions, and genuine independence from the companies being asked to report. None of those exist here. Until they do, passing on this one is the right call. Endorse the goal if you believe in it. Don't confuse the goal with this particular vehicle. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

The past 48 hours have been unusually rich for anyone paying attention to where AI is actually going. Not the headline hype, but the operational reality. Compute budgets revealed under oath. Major employers cutting headcount explicitly because of AI. Government security reviews of unreleased models. An AML agent heading into bank compliance rooms. Two threads running in parallel: massive capital going in, and serious workforce and governance restructuring coming out. Here's what happened. OpenAI Just Put Its $50 Billion Compute Bill on the Public Record OpenAI president Greg Brockman took the witness stand again Tuesday in the ongoing trial with Elon Musk, and the most significant number he delivered wasn't about damages or ownership, it was about spending. Brockman told the court that OpenAI expects to spend $50 billion on computing power in 2026 alone. For context, the company's entire compute budget in 2017 was roughly $30 million. That 1,600x increase in nine years tells you something important about the economics of frontier AI that no press release ever quite captures. This isn't a software business with typical margins. It's a compute-intensive infrastructure play that burns capital at a scale previously associated with semiconductor fabs and satellite networks. Bloomberg and Reuters both confirmed the figures. The trial itself — Musk is seeking over $100 billion in damages — is now functioning as an involuntary transparency event for OpenAI's finances. For enterprise teams evaluating AI vendors: the compute cost picture matters for long-term pricing stability and model availability. Companies spending at this scale need revenue to match, and that eventually flows through to enterprise contracts. Coinbase Cut 700 Jobs and PayPal Followed Within Hours Tuesday was a hard day for fintech headcount. Coinbase announced it would lay off approximately 700 employees — about 14% of its roughly 5,000-person workforce — with CEO Brian Armstrong framing the move explicitly around AI. His language was notable: he described the goal as building "AI-native talent who can manage fleets of agents" and experimenting with "one-person teams" where a single employee combines engineer, designer, and product manager roles. He used the phrase "rebuilding Coinbase as an intelligence, with humans around the edge aligning it" — one of the more explicit statements yet from a major tech CEO about treating the company itself as an AI system. Within hours, PayPal's new CEO signaled plans to cut roughly 20% of the workforce over the next two to three years, also citing AI-driven efficiency. Block, which cut 50% of its staff in February under Jack Dorsey citing secular AI change, has since seen its stock rise about 38%. Coinbase fell 2.6% on the news; PayPal dropped as much as 12%. The pattern across crypto and fintech is now consistent enough to stop calling it coincidental. Block, Gemini, Crypto.com, Coinbase, PayPal — all within roughly the same window, all citing AI. The honest read: some of these cuts are legitimately AI-driven productivity gains, and some are using AI as cover for cuts that would have happened anyway. The question worth asking before your next headcount planning cycle is which category your own efficiency story falls into — and whether your metrics can actually support that answer. FIS and Anthropic Are Deploying an AI Agent Into Bank Compliance Rooms On May 4, FIS — the financial technology company that processes transactions for roughly 12% of the global economy — announced a partnership with Anthropic to build a Financial Crimes AI Agent targeting anti-money laundering operations. The agent is designed to compress AML investigations from hours to minutes by automatically assembling evidence across a bank's core systems, transaction history, and customer activity. BMO and Amalgamated Bank are currently in development with the tool, and broader availability to FIS clients is planned for the second half of 2026. Anthropic's Applied AI team embedded forward-deployed engineers directly inside FIS to co-design the agent and build the evaluation frameworks — a model that's increasingly common when AI moves into regulated, high-stakes environments where getting the architecture wrong isn't just a product problem, it's a legal one. The governance layer here is worth noting: every conclusion the agent reaches links back to source data, and every decision stays with the human investigator. That's not a feature — it's a regulatory necessity. If your team is evaluating AI agents for compliance, audit, or risk functions, the FIS/Anthropic architecture is a reasonable reference model for what "human-in-the-loop in a regulated setting" actually looks like in practice. Microsoft, Google, and xAI Opened Their Pre-Release Models to Government Security Teams The Department of Commerce's Center for AI Standards and Innovation announced on Tuesday that Microsoft, Google, and xAI have signed agreements to give the federal government early access to their AI models — before public release — for national security testing. The trigger was Anthropic's Mythos model, which pushed cybersecurity concerns about advanced AI capabilities to a level that prompted White House consideration of a formal pre-launch review process for frontier models. The deal allows CAISI to probe the models with reduced or even disabled safeguards in order to assess national security-related capabilities and risks. This follows a separate announcement from May 1 in which the Pentagon formalized AI deployment deals with seven companies — Google, Microsoft, Amazon Web Services, Nvidia, OpenAI, Reflection AI, and SpaceX — for use on classified networks. Anthropic is conspicuously absent from both lists after refusing to allow its models to be used for mass surveillance or autonomous weapons. The implication for enterprise AI procurement teams is worth tracking: the government review process that's forming here will likely become a de facto evaluation framework that influences how regulated industries assess frontier models. What CAISI finds in these reviews may eventually surface as procurement guidance or compliance requirements. McKinsey Is Using AI Agents to Staff Its Own Client Teams McKinsey, which has grown to nearly 40,000 employees, announced plans to deploy AI agents to assist in matching consultants to client assignments — a role previously handled entirely by professional development employees who've done this work behind the scenes for decades. The rollout is starting in Latin America and North America, with global expansion planned by end of summer. The goal, per Chief People Officer Wendy Miller, is for professional development employees to spend less time on administrative matching work and more time on counseling and coaching — a genuine shift in the role rather than an elimination of it. McKinsey is also one of the firms that deployed 25,000 internal AI agents across its operations and previously cut around 200 internal tech and support employees after automating non-client-facing work. The staffing question this raises for other professional services firms is straightforward: if the firm that advises companies on organizational design is using AI to manage its own talent deployment, the conversation about AI's role in knowledge work staffing is no longer theoretical. The timeline matters too — a global rollout by end of summer 2026 is a tight implementation window for a system that affects every client engagement the firm runs. Jamie Dimon Put a Number on What AI Is Worth JPMorgan Chase CEO Jamie Dimon made his position explicit on Tuesday: AI will ultimately be worth more than the $1 trillion the industry is on track to invest in it. Dimon has been one of the more consistent voices arguing that AI's value to financial services is underestimated rather than overhyped — a view shaped by JPMorgan's own internal deployment across trading, risk, and operations. That's a notable framing given the current environment. With Big Tech's combined AI capital expenditure for 2026 running toward $725 billion across just four major hyperscalers, and OpenAI's compute bill hitting $50 billion on its own, the question of whether the returns will justify the investment is the defining business question of this cycle. Dimon's view is essentially that they will — and that the companies positioned to capture that value are the ones building now, not waiting for the cost curve to fall. Enterprise Leaders Say AI Is Delivering Value — and AI Accountability Is Now the Biggest Blocker The Jitterbit 2026 AI Automation Benchmark Report, released Tuesday, offered a useful data point for anyone tracking enterprise AI maturity. According to the report, 78% of AI projects are now delivering real business value — effectively declaring the agentic pilot era over. The bottleneck has shifted. The biggest obstacle to scaling is no longer the CFO; it's the CISO. Forty-seven percent of respondents identified "AI accountability" — encompassing security, auditability, and guardrails — as the single most important factor when evaluating new tools. Agent sprawl and what the report calls "agent contamination" are named as real threats to enterprise deployments. A separate Mayfield survey of 266 CIOs, CTOs, and CISOs found that over 72% of enterprises are either in production with or actively piloting agentic AI, with security and risk controls cited as the most frequently noted obstacle to full-scale deployment. The practical read: if your agentic AI initiatives are stalling, the delay is probably governance, not capability. Teams that build the audit trail and accountability layer into their agent architecture from the start — not as a retrofit — are moving faster to full deployment. The FIS/Anthropic model mentioned earlier is one example of how that gets done in practice. Getting your CISO aligned early, with a clear data lineage and override model, is now the unlock — not getting a bigger model or a faster pipeline. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

Today's cycle was heavy with money, governance, and a few reminders that the world is still figuring out who controls what in AI. There were seven meaningful developments — some quiet, some loud — and together they tell you more about where this industry is actually headed than any single headline. DeepSeek's First Outside Money Comes With a Government Tag The most significant news of the day dropped this morning: DeepSeek, China's celebrated AI startup, is raising external capital for the first time. The round, reported by both Reuters and the Wall Street Journal, would value the company at up to $50 billion — up sharply from a $10–30 billion range that was floated just weeks ago. The raise itself is projected at $3 to $4 billion. Two names are in the mix. China's National Artificial Intelligence Industry Investment Fund — the state-backed "Big Fund" created a year ago with roughly $8.8 billion in capital — is expected to lead. Tencent is also in discussions to participate. The implications are worth sitting with. DeepSeek built its reputation by releasing competitive models at a fraction of the cost of its U.S. rivals, and doing so without meaningful external funding. That story is now changing. The government's entry as a lead investor means DeepSeek's development trajectory will increasingly align with Beijing's strategic priorities around AI self-sufficiency and counter-positioning against U.S. export controls. Whether the research stays as open as it has been is a question nobody's answered yet. For teams benchmarking their AI model options, the DeepSeek cost advantage remains real — but understanding the provenance of the models you use is now a due diligence question, not just a philosophical one. Washington Is Eyeing the Shutter Before the Launch The White House is considering an executive order that would require a government review of major AI models before they're made public. Both the New York Times and Bloomberg confirmed the reporting, citing U.S. officials and people briefed on the discussions. The mechanism being explored is a working group of industry executives and government officials. One proposal would give the government first access to new models — not to block them, but to assess potential risks — before public release. White House officials reportedly briefed leadership at Anthropic, Google, and OpenAI on some version of these plans during meetings last week. This is a meaningful shift. The Trump administration came in with a posture of stepping back from AI safety regulation. The Mythos model launch appears to have changed the calculation — specifically, fear of political exposure if a major AI-enabled cyberattack occurred and the government had no prior visibility. The Politico reporting adds that officials are worried about "escalating security risks from advanced artificial intelligence." The practical question for enterprise teams is whether this review process would materially slow model releases, and how that affects vendor timelines. If a working group is created and given real authority, procurement cycles tied to new model launches could get longer and less predictable. Europe Is Asking Anthropic to Test Its Banks While Washington drafts policy, Brussels is making direct calls. EU Economy Commissioner Valdis Dombrovskis confirmed Tuesday that the European Union is in active talks with Anthropic about having EU companies and banks tested for vulnerabilities that Mythos can identify. Speaking after a Eurogroup finance ministers meeting in Brussels, Dombrovskis confirmed: "Indeed there are contacts with Anthropic." Spain's Economy Minister Carlos Cuerpo went further, warning that Mythos-class models may be capable of finding "vulnerabilities or backdoors in virtually all our institutions — not only in the financial sector and companies, but across all sectors." He called for the EU AI Act to be considered as a legislative instrument in response. The framing matters. European officials aren't positioning Mythos as a tool to use — they're treating it as a threat to stress-test against. That's a governance posture, not a procurement posture, and it's a sharp contrast to the U.S. approach of keeping industry in the room. If Anthropic enters formal testing partnerships with EU financial regulators, expect that to create both liability obligations and market access advantages simultaneously. Flex Bets on a Separate Infrastructure Future On the supply side of AI compute, contract manufacturer Flex announced Tuesday that its board has unanimously approved spinning off its Cloud and Power Infrastructure segment into an independent publicly traded company, targeting completion by early 2027. The announcement, confirmed by Reuters and Flex's own press release, frames the move as a way to let each business focus — Flex's core electronics manufacturing on one track, the power and cloud infrastructure build-out on another. The infrastructure unit serves AI data center customers who need both the physical hardware and the power supply to run it. This is the third major infrastructure separation move in recent weeks, and the pattern is becoming clear: companies that built diversified operations are now creating pure-play AI infrastructure vehicles specifically to capture the premium valuations those assets command. Investors want clean exposure, and boards are delivering it. The Flex spin-off joins a crowded field — which is exactly why the next story matters. Wall Street Is Lining Up a $7 Billion Data Center IPO Wave Bloomberg reported this morning that Wall Street banks are preparing to take multiple data center companies public in what could amount to billions in new listings. The Blackstone data-center acquisition vehicle is set to open the sequence next week. DayOne Data Centers, based in Singapore, is in the queue behind it. Together, those two raises could approach $7 billion. There's more behind them. Brookfield Infrastructure Partners-backed CSquare has filed confidentially, and roughly six other companies are circling U.S. IPOs, according to Bloomberg's sources. The volume signals something beyond investor enthusiasm — it reflects a structural bet that AI compute demand will require purpose-built data center capacity at a scale that existing hyperscalers can't fully absorb. The question worth asking is whether the IPO wave front-runs actual demand, or whether it's a rational response to signed contracts that aren't yet public. History suggests it's some of both, and the ones that survive the cycle will be the ones with long-term power agreements and anchor tenants already locked in. Google DeepMind Goes to Space to Train Its Models The strangest story of the day is also one of the more technically interesting ones. Google DeepMind announced Wednesday that it is taking a minority stake — "in the millions" of dollars, per the company's CEO — in Fenris Creations, the newly rebranded studio behind EVE Online. As part of a research partnership, DeepMind will train its models on an offline version of EVE, a massively multiplayer space simulation that has been running for more than two decades. The rationale, from DeepMind's Adrian Bolton, is that EVE Online requires capabilities AI has not yet mastered: long-term planning, continual learning, and operating in a player-driven environment that evolves constantly without a fixed endpoint. DeepMind has previously trained models on arcade classics and StarCraft II, but EVE represents a qualitative step up in complexity — it's an economy, a political system, and a combat simulator all at once. Fenris Creations reported over $70 million in revenue in 2025, making this a healthy company getting a research partner, not a distressed acquisition. For enterprises watching AI capability development, this is a useful signal: the frontier labs are now actively seeking environments that stress-test planning and adaptation at timescales and complexity levels that standard benchmarks don't capture. CDW's Q1 Beat Confirms the Spending Signal A quieter confirmation arrived this morning from CDW, the IT solutions distributor, which posted first-quarter revenue that beat analyst expectations. Reuters reported the results, attributing the outperformance to strong IT demand driven by AI and cloud adoption. CDW doesn't build AI models or run data centers — it's the company that sells, deploys, and services the infrastructure across enterprise customers. A revenue beat at CDW is a clean read on whether mid-market and large enterprise customers are actually opening their wallets, not just expressing intent. They are. For operations and IT leaders, this is useful baseline data. The organizations signaling serious AI investment in earnings calls are backing it with actual purchasing. If you're benchmarking your own AI infrastructure spend against peers, the CDW numbers suggest the pull-forward is real and not concentrated in just the largest tech firms — it's distributed across the customer base CDW serves, which skews toward mid-market and government. The infrastructure is getting built, the money is moving, and the governance frameworks are chasing both. The EU's move to stress-test its banks against Mythos-class vulnerabilities is the leading edge of what every major economy will eventually be doing — not asking whether to regulate frontier AI, but deciding how quickly they can build the institutional capacity to assess it. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

Seventy-nine percent of enterprises have adopted AI agents in some form. Only 11% have them running in production. That spread tells you something important: most organizations have started, and most have not figured out what comes after starting. At a glance OutSystems surveyed enterprise leaders across industries and found 97% expect a material security or fraud incident from AI agents within 12 months. Six percent of security budgets are currently allocated to agent risk. Multi-agent workflow usage grew 327% in under five months across 20,000-plus organizations, including 60% of the Fortune 500 on the Databricks platform. The design pattern reaching production uses a supervisor agent directing specialized subagents, each with bounded scope and access. GPU lead times — the wait between ordering AI processing chips and actually receiving them — stretched to 36-52 weeks in 2026. Enterprises, cloud providers, and AI vendors are all adjusting strategy as a result. NVIDIA's message at GTC 2026 moved from training scale to inference efficiency. That shift aligns more closely with what enterprises actually need to run agents in production. AI-attributed workforce restructuring is running at roughly 16,000 jobs per month in net drag across major technology and services companies. The organizations managing this responsibly are redesigning roles around oversight and judgment. The deployment gap is a governance problem, not a technology problem The OutSystems 2026 State of AI Development report surveyed leaders across industries. Ninety-seven percent of business leaders expect a material security or fraud incident tied to AI agents within the next 12 months. Six percent of security budgets are currently allocated to agent risk. OutSystems responded to their findings by releasing an "Agentic Systems Engineering" framework, defining agent ownership, access boundaries, escalation paths, and audit requirements as core engineering concerns rather than compliance afterthoughts. Whether vendor-published frameworks drive real organizational change is a fair question. The more concrete signal is that major enterprise software companies are now shipping governance frameworks alongside their agent tooling, and that wasn't the case two years ago. The underlying pattern isn't unusual in enterprise technology. New capability spreads quickly. Governance infrastructure lags until something breaks publicly. What's different with AI agents is the consequence window. An unsupervised agent with access to production systems, customer data, or financial workflows can cause significant damage. What production-grade deployment actually looks like Databricks analyzed behavior across more than 20,000 organizations, including 60% of the Fortune 500 on its platform, and found multi-agent workflow usage grew 327% between June and October 2025. Growth continued into 2026. The design pattern reaching production is what Databricks calls the Supervisor Agent structure: one orchestrating agent directing specialized subagents, each with a defined scope and access boundary. For professionals working outside AI architecture, the useful analogy is a department. A manager holds accountability for outcomes. Each team member has a bounded function. When something fails, there is a clear place to look. The structure is auditable in ways a single large agent handling everything is not. The enterprises closing the gap between pilot and production share a consistent trait. They define agent ownership, data access boundaries, human checkpoint requirements, and failure visibility before deployment. These are governance decisions. They don't require more tooling. They require clarity about who is accountable for what. Infrastructure pressure is forcing useful discipline GPU lead times — the wait between ordering AI processing chips and actually receiving them — stretched to 36-52 weeks in 2026. This affects the full chain: cloud providers like AWS, Google Cloud, and Azure building out AI capacity; AI vendors like NVIDIA competing for advanced packaging and memory; and enterprises that depend on cloud-based compute or run their own hardware. Supply cannot match demand, and the organizations feeling it most are the ones that planned to scale through raw acquisition. The shift underway is toward efficiency: heterogeneous hardware configurations, algorithmic improvements, and multi-cloud architectures rather than simply adding more. When compute is scarce and expensive, teams direct it toward agents with measurable business outcomes and cut experiments that haven't produced results. That discipline tends to get deprioritized when capacity feels abundant. The shortage is making it mandatory. NVIDIA and Google Cloud are building toward what comes after the current crunch. At Google Cloud Next 2026, the two companies announced expanded AI Hypercomputer capabilities, with new A5X bare-metal instances running on NVIDIA Vera Rubin NVL72 rack-scale systems. NVIDIA provides the silicon; Google Cloud provides the deployment platform. Together, they're targeting multi-agent and multimodal workloads at production scale, as well as robotics and digital twin applications, which signals that agent deployments are beginning to extend beyond software into systems that interact with physical environments. NVIDIA's message at GTC 2026 reinforced the same direction. The emphasis shifted from training scale to inference efficiency as a first-class design principle. Training is how an AI model gets built, a process run primarily by AI companies and large cloud providers. Inference is how it runs in production every time a person or system uses it, and that's what enterprises manage. NVIDIA moving inference efficiency to the center of its product narrative means the hardware roadmap is converging on what enterprise deployments actually require. Power is the constraint that's harder to engineer around quickly. Most enterprises access AI compute through cloud providers, so the energy problem is one step removed, but it shows up in data center availability, cost, and capacity constraints in major markets. Meta, which builds and operates its own AI infrastructure at scale, has reportedly been exploring space-based solar to address its energy requirements. Space solar is a decade-scale research initiative. The fact that a company with Meta's resources is evaluating it is a concrete indicator of how far outside normal grid planning the AI infrastructure problem has extended. For enterprises, the near-term reality is constrained capacity and elevated costs. New agent deployments should be costed at production volume, not at proof-of-concept scale. The workforce transition and what responsible looks like AI-attributed workforce restructuring is running at roughly 16,000 jobs per month in net drag across major technology and services companies. Agents and automation are being cited explicitly in earnings calls and restructuring communications across software, financial services, and professional services. The honest version of that trend: contact center roles, routine data processing, and first-tier support functions are being reduced. Agents handle high-volume, pattern-matching work consistently and at lower cost. Companies deploying agents at scale will need fewer people for those functions. The data across OutSystems, Databricks, and current earnings commentary all point in the same direction. The disruption isn't limited to frontline roles. On May 4, Anthropic announced a $1.5 billion joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs to launch a new AI-native enterprise services firm. The company will deploy Claude into core business operations for midsize companies, offering the kind of strategic transformation work that management consultants have historically delivered. OpenAI is reportedly pursuing a near-identical structure with TPG and Bain Capital. Business Insider, citing a source with direct knowledge of the deal, described the Anthropic venture as the McKinsey of AI. That framing is worth sitting with. The workforce disruption from AI isn't only happening inside companies. It's reorganizing the service industries that advise them. If AI-native firms can deliver transformation outcomes at lower cost and greater speed, the consulting model faces the same pressure that contact centers do. The distinction that matters for organizations and their people is whether roles are being eliminated or redesigned. Contact centers are the clearest current example. The organizations handling this well are shifting workers from routine inquiry handling to reviewing edge cases, correcting model errors, managing escalations, and maintaining the oversight layer that agents cannot supply on their own. The ratio of humans to customer interactions changes. The nature of the human role changes from pattern-matching to judgment. What those redesigned roles look like in practice: Reviewing cases flagged for escalation, rather than handling every incoming interaction Identifying patterns in agent errors that reveal process or data quality problems Managing the configuration and rules that govern agent behavior Handling exceptions where customer history, context, and judgment change the right answer That redesign path is not only the more responsible workforce approach. It's the more durable business strategy. The governance gap in the OutSystems data, with 11% of agents in production and 6% of security budgets allocated to agent risk, is exactly the gap filled by people who understand both the operational work and the AI systems handling it. That gap doesn't close by eliminating the people closest to the operations. It closes by repositioning them as the oversight layer that enterprise agents actually require. The companies treating AI adoption as primarily a headcount reduction exercise are optimizing for near-term margin. The ones building human oversight into their agent architecture from the start are building something that survives the first material security incident. Based on the 97% figure, that incident is coming for most of them within 12 months. Whether the oversight structure is already in place when it does is the question worth asking now. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

At a glance Anthropic is in talks on a $200M joint venture with Blackstone and Hellman & Friedman to embed Claude across hundreds of PE portfolio companies, replacing the software and advisory services those companies currently buy KPMG cut approximately 400 US advisory jobs this week, roughly 4% of its US workforce, citing AI automation and slowing demand for traditional consulting services Accenture invested in General Robotics to deploy physical AI (AI combined with physical robotics systems) through the GRID platform for manufacturers and logistics operators, positioning inside NVIDIA's physical AI ecosystem Salesforce reports AI agents now handle 50% of its customer service interactions, which the company ties directly to its headcount rebalancing strategy Meta confirmed 8,000 job cuts beginning May 20 and the closure of 6,000 open roles; Microsoft offered voluntary buyouts to roughly 7% of its US staff simultaneously Four of the largest consulting and professional services firms in the world made significant AI positioning moves this week. They chose very different approaches. Meanwhile, the most aggressive play of the group came from a company that isn't a consulting firm at all. The pattern is worth understanding before it becomes your competitive context. Anthropic Is Building the Service Layer PE Firms Used to Buy Elsewhere Anthropic is in advanced talks to invest $200 million in a joint venture with private equity firms Blackstone and Hellman & Friedman, according to the Wall Street Journal. The structure is deliberate: the JV would sell Claude-powered AI tools alongside consulting and integration services directly to the portfolio companies owned by those PE firms. The model mirrors Palantir's enterprise deployment approach, where AI capability is sold directly into client operations rather than licensed as software for clients to use independently. The scale potential is not abstract. Blackstone and Hellman & Friedman collectively control hundreds of portfolio companies across industries. The JV gives Anthropic a direct sales channel into that entire base, with PE firms acting as the distribution layer. For portfolio companies, the pitch is AI services that replace legacy software subscriptions and advisory hours from third-party vendors. That last sentence is the one to sit with. The firms currently selling those software subscriptions and advisory hours include some of the largest names in professional services. Notably, the earlier reporting on May 4 covered a combined $11.5 billion in joint ventures between OpenAI and Anthropic separately. The $200 million Anthropic figure is the specific investment amount within the Blackstone and Hellman & Friedman structure and includes the consulting deployment component, which is the more consequential detail from an industry disruption standpoint. KPMG Put a Headcount Number on the Demand Shift KPMG cut approximately 400 US advisory positions this week, roughly 4% of its US workforce. The company cited two contributing factors: AI automation reducing demand for traditional advisory hours, and an overall slowdown in consulting demand. Both pressures are self-reinforcing. As AI handles more of the analysis, synthesis, and compliance work that advisory teams have historically billed for, clients need fewer human hours per engagement. As the billable hour volume drops, revenue per client relationship compresses. That dynamic doesn't reverse when demand recovers because the AI capability doesn't go away. For leaders managing professional services teams, the useful exercise is sorting current deliverables by type: analysis and synthesis tasks (where AI performs well) versus judgment, negotiation, and relationship-dependent work (where human performance still dominates). The mix shifted meaningfully in the past two years. The KPMG cuts suggest the revenue math has followed. Accenture Placed Its Bet on Physical AI Accenture announced an investment in General Robotics to build out what the company calls physical AI: the combination of AI decision-making with physical robotics systems operating in real-world environments like factory floors and warehouse logistics. The partnership is built around General Robotics' GRID platform, a unified intelligence layer that connects robots across different manufacturers and allows them to be deployed as coordinated, continuously adapting systems. Accenture described the investment as extending its enterprise orchestrator role inside NVIDIA's physical AI ecosystem. The near-term focus is manufacturing and logistics: autonomous operations for asset-intensive industries where robot deployment at scale has historically been expensive and fragile. The strategic logic is distinct from the software-services model. Accenture is betting its next wave of services revenue comes from deploying and orchestrating hardware-plus-AI systems that its clients cannot build or manage themselves. That's structurally harder to displace with a software-only JV. An AI language model can generate an analysis; it can't replace a robot running a conveyor system calibrated to client-specific specs. Bain Bought Startup Access Instead Bain & Company took a different path. The firm formalized partnerships with seven venture capital firms through its Venture Ecosystem team, giving clients direct co-innovation access to AI startups and early-stage AI capabilities. Bain did not disclose specific VC firm names or investment amounts in the announcement. The strategic logic is a lighter-capital bet than Accenture's hardware play: Bain is positioning as the bridge between established enterprise clients and the AI startup ecosystem, rather than acquiring or building AI capability directly. That approach works if clients continue to value the bridge and don't develop their own direct VC relationships. It's a model that depends on ongoing client trust in Bain as the filter. That's a reasonable bet for a firm with Bain's relationship depth, but it's more exposed to disintermediation over time than a capability-based play. Salesforce Showed What 50% Displacement Looks Like in Practice Salesforce reported this week that AI agents now handle 50% of its customer service interactions. The company has explicitly connected this figure to its headcount rebalancing approach, using AI-handled volume to offset human staffing requirements rather than growing both in parallel. The customer service context matters. First-line customer support has been the largest category of AI-related displacement so far, not engineering, not finance, not strategy. Salesforce, which both uses enterprise AI internally and sells it to others, is one of the clearest examples of automation economics at operational scale: when the tool is reliable and the volume is sufficient, the headcount math changes. The 50% figure also serves as a reference point for what "headcount rebalancing" actually means in numerical terms. If half the interaction volume moves to AI, the human team doesn't drop by half (relationship escalations, complex cases, and quality oversight still require people), but the required headcount for the total function is substantially lower. Meta and Microsoft Made Their Cuts Concrete Meta confirmed 8,000 job cuts beginning May 20, alongside closing approximately 6,000 open roles that were already in the hiring pipeline. Microsoft simultaneously announced voluntary retirement buyouts targeting roughly 7% of its US workforce of approximately 125,000 employees. On Meta's January earnings call, Mark Zuckerberg called 2026 "the year that AI starts to dramatically change the way we work." The May 20 start date makes that a schedule, not a prediction. Combined with Amazon's 16,000 layoffs earlier this year, the pattern at major tech companies is consistent: AI infrastructure budgets are increasing while total headcount decreases. These are not restructurings driven by revenue decline. Meta, Microsoft, and Amazon are all growing. The workforce reductions are an explicit trade against AI-handled capacity, which is a structural claim about where production value is being created. Enterprise Agents Are Deployed Faster Than They Are Governed Deloitte's 2026 State of AI in the Enterprise data shows that 21% of companies have a mature governance model for their AI agents. Multiple vendors including Zenity, Witness.ai, DataRobot, and Palo Alto Networks have released practical governance frameworks for autonomous agents (frameworks covering how agents are authorized to act, what data they can access, and what real-world tasks require human sign-off before execution), but no single standard has been adopted at scale. The practical result is shadow adoption: agents running in production workflows without clearly defined authority boundaries, escalation paths, or audit trails. Deployment is moving faster than the control layer. That gap has existed for two years and is widening as the number of deployed agents increases. The governance problem isn't that enterprises don't care. It's that the economics push toward deployment before the controls are ready. Slowing deployment to build governance first has a cost that is visible and immediate; an ungoverned agent causing an error is hypothetical until it happens. Most teams are making a reasonable short-term calculation that creates a long-term exposure. The week's broader picture is a single professional services economy in motion simultaneously. One firm is cutting advisory headcount. Another is embedding startup ecosystems. A third is going physical. And an AI company is going straight at the consulting margin itself. There is no single right answer in that mix, but there is a clear wrong one: staying still while the structure shifts. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

At a glance: Cloudflare cut 1,100 employees — 20% of its workforce — after internal AI usage surged more than 600% Oracle reportedly eliminated 20,000–30,000 positions in 2026, redirecting spend to AI infrastructure and data centers IBM confirmed it replaced approximately 200 HR professionals with AI agents this month Perceptyx research: trust in leadership is a stronger predictor of AI adoption success than technical skills or training McKinsey: 70% of the effort in a successful AI transformation belongs to people and process, not the technology Ciena deployed MoveWorks agentic AI across 100+ IT and HR workflows; approval times dropped from days to minutes Salesforce launched "Managing at Salesforce" — a formal program teaching managers to lead teams of humans and AI agents together; 79% of workers expect to need reskilling The past two weeks moved the workforce conversation from projection to headcount. Specific companies. Clear numbers. Stated triggers. The pilot era is over for any organization paying attention. Cloudflare's Internal Math Cloudflare announced cuts of more than 1,100 employees — roughly 20% of its total workforce — as it reorganizes around what the company calls the "agentic AI era." The deciding factor wasn't a revenue shortfall or a market downturn. Internal AI usage at Cloudflare surged more than 600% in recent months, and the operational and back-office functions that previously required human bandwidth simply require less of it now. That 600% usage figure is the telling detail. When you see that kind of internal adoption rate, you're not looking at a cost-cutting story with AI as convenient cover — you're looking at a company that deployed its own tooling at scale, watched the workload shift, and adjusted staffing to match reality. That's a different animal from a restructuring dressed up in AI language. For context: more than 90,000 tech-sector jobs have been eliminated across the industry in 2026 so far. Customer support, software development, data analysis, HR, and admin functions are showing the highest concentrations. Cloudflare's cut stands out because the trigger is specific and public — which makes it a useful data point for anyone trying to track how agentic deployment actually translates into workforce decisions. Oracle's Infrastructure Trade Oracle reportedly cut between 20,000 and 30,000 jobs this year. The stated direction: redirect that budget into AI infrastructure, cloud computing expansion, and data center development. Head count out, compute capacity in. The pattern is consistent with what's visible across several large enterprise software vendors. As AI-augmented operations require less human coordination of routine processes, the spend calculus shifts toward the infrastructure that powers those operations. Oracle is a company with a substantial base of traditional enterprise software and services revenue, and the transition isn't clean — some of these reductions may be tied to broader competitive repositioning rather than AI displacement alone. The attribution question in large-enterprise restructuring is always messy. What's not messy is the investment direction. Oracle isn't cutting and sitting still. It's explicitly trading operational headcount for compute and AI capability, which is a clear signal about where leadership believes the value will be generated going forward. IBM Replaces 200 HR Professionals with AI Agents This is the sharpest example of the week. IBM confirmed it replaced approximately 200 HR professionals with AI agents this month, covering recruiting coordination, onboarding, benefits administration, and performance support. These are structured, process-heavy functions — they follow repeatable workflows, answer predictable questions, and handle case management that fits well within what current AI agents handle reliably. IBM is not a startup running a pilot. It is one of the largest enterprise technology companies in the world, and it made a 200-person HR reduction in a single move. Similar, quieter restructuring has been reported at Moderna and at federal agencies managing HR functions at scale. For anyone leading an HR function today: if your team spends significant time on tier-one support, routine administration, or templated communications, that work is structurally exposed. The question isn't whether AI can handle it — IBM answered that. The question is what your team reorients toward, and how fast. The organizations that get ahead of this are already identifying which HR work requires human judgment, empathy, or complex stakeholder management — and building those capabilities now rather than defending the workload that's already leaving. The Invisible Factor in AI Adoption While the headcount numbers get the headlines, a Perceptyx research report this week offers a sharply different angle: the strongest predictor of successful generative AI adoption is not technical training, tool access, or employee skill level. It's trust. Employees with high trust in their organization and leadership are more likely to view AI-driven change as an opportunity. Employees with low trust resist the tools even when the tools demonstrably work. The research also surfaces a pattern called "AI angst" — a state where fear of replacement paradoxically increases tool usage while simultaneously raising resistance to the broader transformation. Employees use the tools to protect their relevance while quietly opposing the change around them. That dynamic has real consequences for leaders managing adoption. You can invest substantially in licenses, training programs, and change communications. If the underlying trust relationship with your workforce is broken or damaged, adoption rates will underperform relative to organizations that addressed that problem first. The frameworks emerging from this research focus on three levers: genuinely accessible tools (not just technically available ones), workflows that position AI as augmentation rather than substitution, and explicit organizational legitimation — meaning leaders visibly using the technology and publicly endorsing new ways of working. None of those levers is primarily a technology decision. Where the Work Actually Lives McKinsey's QuantumBlack team published updated guidance this week for enterprises that have moved past early pilots and are now reconfiguring operations around generative AI. The core finding: 70% of the effort in a successful AI transformation belongs to people, process, and change management. The technology is the smaller part of the work. The leading organizations they describe are not running AI alongside existing processes. They are redesigning operating models around it — restructuring workflows, rebuilding managerial accountability, and treating change management as an executive competency rather than a communications task. That reframing matters. Traditional change management centered on explanation: telling people what was changing, giving them time to adjust, answering questions. What McKinsey is now describing is closer to organizational design — active sponsorship, reinforcement at every management layer, and deliberate capability-building. The managers who internalize that distinction now will be significantly better positioned than those still treating AI transformation as a software rollout with a training module attached. Agentic AI in Production: Ciena's 100-Workflow Deployment Ciena, the networking and software company, deployed MoveWorks' agentic platform across more than 100 IT and HR workflows. The agents handle diagnostic tasks, credential resets, cache clearing, and multi-step approval processes end-to-end — without requiring a human at each decision point. Approval times dropped from days to minutes. Ticket volume across the affected functions fell materially, though MoveWorks has not published an independent breakdown of the specific reduction figures. The capability here is worth being precise about. These are not chatbots or rules-based automation. Agentic systems like the ones Ciena deployed can orchestrate across multiple platforms and data sources, handle branching logic, and complete multi-step tasks without constant prompting. That's a different capability class from workflow automation tools that organizations have used for years. The functions they replaced — service desk work, IT support coordination, HR case routing — were reactive, structured, and repetitive. That profile is the highest-risk category for agentic displacement across enterprise functions. Salesforce Teaches Managers to Lead Agents Salesforce launched an internal program called "Managing at Salesforce" to teach its managers how to lead what the company describes as the digital labor era — teams composed of both humans and AI agents working alongside each other. The fact that Salesforce needs a formal program to teach this is worth sitting with. This is a company whose entire commercial pitch is built on AI agents. If its own managers require deliberate training to operate effectively in that environment, it is a candid acknowledgment that the behavioral and leadership skills needed for human-agent collaboration don't emerge from exposure alone. They have to be built. A separate survey cited this week found that 79% of workers expect to need significant reskilling because of AI. The leaders most effective at managing through that transition are the ones building managerial capability ahead of full agent deployment, not after. Salesforce building an internal program while selling the same transformation externally is either a reassuring signal that they're eating their own cooking — or an ironic one, depending on how the results look in six months. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

IBM replaced approximately 200 HR professionals with AI agents this month. The full function suite is live — recruiting, onboarding, benefits administration. This wasn't a pilot announcement or an aspirational roadmap slide. Confirmed production deployment in a major enterprise, running now. At the same time, Cloudflare cut 1,100 people — roughly 20% of its workforce — with the company explicitly citing a 600% surge in internal AI usage over recent months as the operational context for the reduction. Oracle is working through a separate reduction of 20,000 to 30,000 jobs, redirecting the headcount spend toward AI infrastructure and data centers. The decisions are made. The transitions are underway. What's less settled is whether any of this goes well. That's not pessimism. It's the consistent signal coming out of a wave of research published this spring. McKinsey, BCG, and Kearney each published independent findings pointing at the same variable: roughly 70% of what determines AI transformation success is people, process, and change management — not the technology itself. The model selection, the compute allocation, the vendor evaluation — all of that matters, but it accounts for the minority of the equation. Most of what determines whether AI deployment turns into actual organizational improvement sits in the management layer. What the Adoption Research Is Actually Showing Perceptyx sharpened that finding further. Their analysis of employee listening data found that trust in leadership is a stronger predictor of generative AI adoption success than technical proficiency or tool access. Not training depth. Not software quality. Trust. That result has direct implications for anyone designing or sponsoring an AI rollout. It suggests the primary bottleneck in most organizations isn't tooling — it's credibility. Employees adopt AI faster and more thoroughly when they believe leadership is being straight with them about what's changing, why, and what it means for their role. Resistance patterns, Perceptyx found, tie more to perceived fairness and the quality of communication than to fear of job loss alone. For companies deploying AI while simultaneously announcing large headcount reductions, that's a complicated starting position. IBM, Cloudflare, and Oracle are doing exactly that. The technology deployment may be technically sound. Whether the remaining workforce absorbs it productively depends on organizational dynamics that most AI vendors don't have a product for. The practical implication is concrete: if your rollout plan allocates most of its management attention to the technology stack and treats communications as a downstream activity, the research suggests you're likely to underperform your implementation investment. The organizations getting consistent results are doing something different. The Change Muscle Problem BCG and Kearney's framing is worth spending time on. Their research describes what they call a "change muscle" — the organizational capability to absorb, adapt to, and continuously integrate new operating models. The companies performing best on AI transformation treat this as a permanent operational capability, not a one-time project. They've structurally embedded continuous transformation into how they run. That's a different management challenge than selecting tools or setting AI adoption targets. It requires investment in change infrastructure — sponsorship structures, feedback mechanisms, reinforcement systems — that most organizations have never built because they didn't need them continuously. They built them for specific initiatives and then wound them down. The Kearney finding is that winding them down is the mistake. The pace of change coming from AI deployment means transformation is no longer episodic. Companies that treat each AI initiative as a discrete project are building and dismantling change capacity repeatedly. The ones doing it well have stopped cycling through that. Where the Production Results Show Up Ciena's deployment of MoveWorks' agentic AI system is one of the cleaner case studies available at this scale. More than 100 workflows automated and running in production. Approval cycle times reduced from days to minutes in enterprise networking operations. Named company, specific outcomes, real production context — not a controlled pilot with carefully selected use cases. What's notable about the Ciena example isn't just the efficiency improvement. It's how the deployment was scoped. Rather than a broad AI rollout across the organization, the implementation targeted defined business processes with measurable output cycles and clear decision points. Approval workflows are a natural fit: the inputs and outputs are structured, the handoffs are documented, and the time from submission to decision is straightforward to track. That scoping approach makes results quantifiable and the business case defensible. It also makes change management more tractable — employees whose approval workflows changed can see the before and after clearly, which reduces the trust deficit that Perceptyx's research identifies as the primary adoption constraint. The Salesforce Indicator Salesforce now reports AI agents handle approximately 50% of customer service interactions. The company has built a formal program — "Managing at Salesforce" — specifically to train managers on leading mixed teams of people and AI agents. That program exists because the challenge is real: managing a hybrid team requires skills that aren't covered by traditional management development. Workflow oversight, output review, escalation protocols, and accountability structures all work differently when some of the team members are automated systems. The 79% figure from Salesforce's workforce data carries weight beyond the company itself. Nearly eight in ten of their workers say they expect employer-supported reskilling as AI scales. That expectation is no longer limited to tech company employees or early adopters. It's spreading across industries as deployment accelerates, and it's shifting from a request into an expectation that affects hiring, retention, and engagement. Organizations treating reskilling as optional or deferred are likely to see the cost show up differently — through attrition among the people they most want to keep, or resistance patterns that slow adoption of the systems they've invested in building. The Strategy Layer Is Restructuring Too What's happening inside companies is being mirrored by structural changes in the advisory industry around them. Anthropic is in active discussions with Blackstone and Hellman & Friedman on a joint venture to embed Claude across their private equity portfolio companies. The model is explicitly Palantir-style: not a software license with an implementation partner, but deployed AI combined with integrated consulting services. The focus areas are financial analysis, software engineering, and customer operations — the same white-collar functions where most enterprise AI adoption is currently concentrated. Blackstone and Hellman & Friedman collectively have hundreds of portfolio companies. Successful execution would move Claude into professional functions at a scale that direct enterprise sales would take years to reach. Bain & Company moved in a different direction toward the same goal. The firm formalized partnerships with seven major VC firms under a Venture Ecosystem program. The design gives Bain clients direct co-innovation access to AI startups — early visibility into what's being built and the option to co-develop before products reach the open market. The structure is about giving clients an advantage in identifying what's coming before competitors do. Neither of these is an announcement about adding AI tools to a consulting practice. Both are structural shifts in how strategy advice gets packaged, delivered, and monetized. Consulting firms are positioning to own the transformation work, not just recommend it. For executive teams working with major advisors, that shift is worth understanding clearly: when the advisory firm's business model includes a stake in the deployment outcome, the nature of the advice changes. It's worth asking what that means for how recommendations get shaped. The Question Under All of This The technology decisions at IBM, Cloudflare, Oracle, Salesforce, and Ciena are largely made. The headcount restructuring is in motion. The PE and consulting firms are moving capital and partnerships to capture implementation value at scale. The consistent finding across McKinsey, BCG, Kearney, and Perceptyx research is that the constraint separating companies that perform from companies that struggle isn't access to technology. It's the organizational infrastructure built around the technology — leadership credibility, continuous change capacity, and genuine investment in the people whose roles are being redesigned. Companies building that infrastructure deliberately are producing results like Ciena's: specific, measurable, repeatable. Companies treating it as the soft layer that follows the real work are accumulating a change management debt that tends to come due right when they need adoption most. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

OpenAI announced yesterday the launch of the OpenAI Deployment Company — a standalone entity backed by $4 billion in committed capital, 19 global institutional partners, and 150 engineers ready to embed inside enterprise organizations from day one. The structure is deliberate. Rather than sell API access and leave implementation to third parties, OpenAI's Deployment Company sends engineers directly into client organizations to redesign critical workflows around AI. This is professional services, not a software subscription. The Forward Deployed Engineer Model The Deployment Company mirrors Palantir's playbook. Forward Deployed Engineers don't work from an OpenAI office. They embed inside the client's operation, map the workflows that matter most, redesign them around AI capability, and stay through deployment into production. To launch with that capability immediately, OpenAI acquired Tomoro, an applied AI consulting firm. Tomoro brought 150 FDEs to the table on day one. Current enterprise clients already include Oracle, State Farm, and Uber. The combination of software and embedded engineers is what makes Palantir's contracts defensible. Software alone can be switched out. Software plus engineers who know the client's systems, processes, and internal politics is a different kind of relationship. OpenAI is making the same structural bet. The Investor List Changes the Story The $4 billion is backed by names you'd expect at this scale: TPG, Advent International, Bain Capital, Brookfield, Goldman Sachs, SoftBank Corp., and Warburg Pincus. Large financial institutions and private equity firms taking positions in the enterprise AI services market. The interesting names are the consulting firms also on the cap table: Bain & Company, Capgemini, and McKinsey & Company. These three firms are not passive investors. They are the organizations that currently charge enterprise clients to do exactly what the OpenAI Deployment Company will do: embed advisors inside organizations and lead AI transformation programs. McKinsey has built a substantial AI practice. Bain has done the same. Their people are sitting in client boardrooms right now advising on the same workflows OpenAI's FDEs will redesign. They chose to invest rather than compete directly. That is a significant strategic signal. One reading: the Deployment Company is complementary. FDEs handle technical implementation while traditional consultants own strategy, change management, and the board relationship. McKinsey advises the CEO on the transformation vision. OpenAI's engineers build the systems underneath it. A second reading: they needed a position. If the Deployment Company succeeds at scale, it will not just take implementation work. It will build client relationships of its own. Investing now gives McKinsey and Bain visibility into the model, potential preferential partnership access, and upside if the venture succeeds. It is a hedge as much as a collaboration. Both can be true at the same time. Distribution First, Service Second Enterprise buyers should understand one structural reality before signing with the Deployment Company: it is majority-controlled by OpenAI. The FDEs who embed inside your organization understand your workflows, your data, and your operational priorities. They also work for OpenAI. The AI systems they design are built around OpenAI's models. This is a distribution mechanism as much as a transformation advisory service. That is not necessarily disqualifying. Palantir's clients understand they are buying the Palantir stack. They accept that constraint because the embedded execution delivers real operational change that would not happen otherwise. Some enterprise buyers will make the same calculation here. The question worth asking before engagement: do you want the model provider also serving as your transformation architect? For organizations that have already built on a different AI stack, or that want model-agnostic advice about where AI should be deployed in their operations, the Deployment Company is a specific kind of partner with a specific set of incentives. Understanding that going in matters. The Competitive Pressure Falls in the Middle The 19-partner consortium is making a long-duration bet: that enterprise demand for embedded AI expertise is large and durable, and that combining proprietary models with embedded professional services creates a defensible market position. Large consulting firms have the assets to absorb this pressure near-term. Board-level relationships, sector expertise, regulatory knowledge, and global scale do not disappear because OpenAI hired 150 engineers. McKinsey and Bain's investment suggests they have already calculated that their core market is defensible, at least for now. The organizations with less buffer are in the middle tier: boutique AI implementation firms, smaller systems integrators, and AI consultancies that compete on technical execution without the relationship assets of a McKinsey or the model advantage of an OpenAI. The Deployment Company's FDEs are entering exactly the market those firms occupy. They will feel this faster. One more variable worth watching: the Deployment Company's thesis partly depends on OpenAI's models remaining best-in-class for enterprise workflows. That may be true today. The competitive dynamics of the AI model market over the past 18 months have not suggested any vendor holds that position permanently. The 19 partners are betting it holds long enough for the services layer to become self-reinforcing. That is not a guaranteed outcome. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together.

The conversation about AI at work has shifted. It's no longer about whether to adopt — it's about who can actually scale. New data and real implementation stories out today show a widening gap between organizations running AI in production and those stuck in pilot purgatory. The specifics are worth your attention. Here's what happened in the last 24 hours. The Scaling Problem Is Real, and C-Suites Are Feeling It Infor's Enterprise AI Adoption Impact Index surveyed 1,000 executives across retail, manufacturing, and logistics. The finding: more than half of businesses are struggling to scale AI beyond initial workflow use cases. That's not a technology problem. That's a process, change management, and integration problem. C-suite leaders in these industries are using AI for workflows — that part's happening. But moving from a handful of power users to organization-wide adoption is where momentum stalls. If your team is in that gap right now, you're not alone, and the issue likely isn't the tool. The question worth asking is whether you've defined what "scaled" actually means for your organization before you started. Most teams haven't. Where Agentic AI Is Actually Starting Inside Enterprises Intellias, citing Deloitte research, offers a clear picture of where enterprises are choosing to start with agentic AI (AI systems that can take multi-step actions autonomously): engineering. Specifically, prototyping, code generation, and testing. That's not a coincidence. Engineering workflows have clear inputs and outputs, measurable quality signals, and teams that can evaluate AI accuracy without being fooled. Starting agentic AI in a function where you can actually verify the output is smart risk management. For CIOs thinking about where to pilot agentic systems, the Deloitte data suggests engineering is the lowest-risk, highest-feedback entry point before expanding to other functions. Once you've built the governance muscle there, you'll have a better framework for the messier domains. Morgan Stanley's Approach to Advisor Productivity Calls9 Insights reports that Morgan Stanley has deployed GenAI to analyze over 1 million annual conference calls, surfacing insights for financial advisors to improve client interactions. The goal: streamline advisor workflows, not replace advisor judgment. That distinction matters. AI handling the information synthesis layer — reviewing transcripts, flagging themes, surfacing patterns — while human advisors apply relationship context and professional judgment is a model that works. It's not glamorous, but it's the right architecture for high-stakes professional roles. If you're in financial services or any domain where client relationships carry real risk, that's the design principle worth borrowing. Translation Work Is Getting Faster, But the Nuance Still Requires Humans TIME's coverage of white-collar AI shifts includes a look at DeepL's new AI translation tool. DeepL CEO Jaroslaw Kutylowski highlights a specific feature: custom glossaries and follow-up questions that let translators work faster in niche domains where precision is non-negotiable. This is a useful example of what good human-AI workflow design looks like. The AI handles volume and consistency; the human expert handles domain accuracy and edge cases. Speed goes up, quality holds, because the tool is built around the human's workflow rather than trying to replace it. The professional translation market is one of the cleaner examples of AI augmentation done right — and the lesson extends well beyond language work. Headcount Planning Is Getting a Real Upgrade ChartHop's AI headcount planning tool now generates multiple hiring or restructuring scenarios in minutes, with instant budget impact modeling and org design visualization. What used to take a people operations team days of spreadsheet work — modeling three to five headcount scenarios against budget constraints — is now a fast iteration loop. This matters most during planning cycles and any period of organizational change. If your team is still doing headcount modeling in static spreadsheets, you're spending time on structure instead of decisions. The practical move: evaluate whether your current planning toolset can run scenario modeling at speed, or whether you're artificially slowing down leadership decisions because of tool friction. IT Leaders Are Running AI Agents in Production — More Than Half of Them Moveworks reports that more than half of IT leaders are now running AI agents in production environments, with Moveworks agents automating both routine and complex tasks across enterprise workplace applications. That's not a pilot stat. That's production. And it marks a meaningful shift from AI as a bolt-on to AI as infrastructure inside IT operations. For business leaders outside IT: this is your best internal case study. Your IT organization has likely already worked through the integration, security review, and change management challenges that your function is still anticipating. Talk to your CIO before you build your own roadmap from scratch. HR Automation Is Covering the Full Employee Lifecycle Leapsome's recent breakdown of HR automation lays out exactly where AI is being applied in HR right now: performance reviews, goal-setting, onboarding, learning pathways, and payroll. Not just one piece — the full employee lifecycle. The implication is straightforward. HR teams that were already under-resourced relative to headcount now have a path to run more consistent, timely processes without adding staff. The risk, as always, is that automation without human oversight produces fast but shallow results — especially in performance and onboarding, where the quality of the experience directly affects retention. Automate the administrative layer. Keep humans in the loop where the experience actually shapes employee trust and performance. Y Combinator Startups Have Made LLMs a Baseline Requirement Bee Techy's 2026 State of Enterprise AI report puts a sharp number on where the startup market has landed: 92% of Y Combinator startups have integrated LLMs into their core product architecture. For venture capital, it's now effectively table stakes. That has downstream effects on enterprise buyers. Products you're evaluating — whether for HR tech, operations, sales, or customer service — are increasingly built on large language model infrastructure from the ground up. That changes the due diligence conversation. You're no longer asking "does this product have an AI feature?" You're asking how the AI layer is built, what data it touches, where it can fail, and who's accountable when it does. Procurement and security teams that haven't updated their vendor review frameworks for LLM-native products are working with the wrong checklist. Consulting Firms Are Using AI to Reimagine Delivery Models SAP's resources on AI in consulting detail how firms are combining AI tools with human expertise to accelerate transformation project delivery. The focus is on faster time-to-value for clients — less time on information gathering and structuring, more time on decisions and implementation. Panorama Consulting Group adds specifics on where this is showing up in professional services: capacity planning, project risk detection, and financial forecasting inside ERP and PSA systems (professional services automation platforms that manage projects, resources, and billing). If you're buying consulting services right now, it's fair to ask your partners how AI is affecting their delivery model — and whether those efficiency gains are being passed to you or absorbed into margins. What Connects All of This The through-line today is execution discipline. AI tools are no longer in short supply. The challenge is building the processes, governance, and human-AI workflow designs that let organizations actually capture value from them at scale. The Infor data says more than half of enterprises are struggling to scale. The Morgan Stanley and Moveworks examples show what working at scale looks like. The gap between those two conditions isn't technology — it's clarity on where humans add irreplaceable value, and building AI workflows that support that rather than work around it. That's the conversation happening inside successful organizations right now. It's worth having on your team before your planning cycle, not after. If you want to stay ahead at the intersection of AI, automation, and human performance — where technology meets psychology, processes, and real workplace behavior — subscribe to Agenticism. We cut through the hype to deliver practical insights for leaders focused on making people, processes, and technology work better together. Sources Calls9 Insights — View Article TIME — View Article Bee Techy — View Article ChartHop — View Article Moveworks — View Article Leapsome — View Article Infor — View Article SAP — View Article Panorama Consulting Group — View Article Intellias — View Article

Four verified moves landed in 24 hours — each one shifting how enterprises buy, govern, and think about AI. OpenAI and Microsoft ended the exclusivity arrangement On April 27, the two companies amended their partnership. Microsoft's license to OpenAI technology is now non-exclusive through 2032. OpenAI can ship models first on Azure but then serve customers on any cloud. Azure stays the primary home, but the lock-in is gone. For enterprise teams this matters immediately. Mixing models across cloud providers without ripping out existing contracts is now possible in a way it wasn't before. The practical question: which of your current AI workflows are tied to one provider because of a contract structure versus because that provider genuinely performs best for that task? The new arrangement gives you leverage to separate those decisions. Google joined the Pentagon's classified AI program April 28 reporting from The Information, WSJ, and Reuters confirmed Google signed a deal giving the Department of Defense access to its AI models on classified networks — joining OpenAI and xAI after Anthropic passed. The agreement covers "any lawful government purpose," with some limitations on autonomous weapons and mass surveillance. The useful signal here isn't about defense contracting specifically — it's about deployment speed. Frontier AI models are now cleared for classified operations at the highest levels of government. If the risk tolerance for AI deployment at that level has shifted, it's worth examining whether your own organization's risk thresholds are calibrated to current reality or to assumptions from 2023. OpenAI missed internal growth targets WSJ reported April 28 that OpenAI fell short on monthly revenue goals and the target of 1 billion weekly active ChatGPT users by end of 2025. Competition from Anthropic in coding and enterprise is real. Internal spending on data centers is raising questions even within the company. This doesn't change the trajectory — OpenAI is still growing fast. What it does puncture is the narrative that AI adoption is frictionless and automatic. The practical implication: build in conservative adoption curves and measure actual utilization, not licensed seats or activated accounts. David Silver raised $1.1 billion to build agents that need less human data On April 27, David Silver — the DeepMind researcher who built AlphaGo — closed a $1.1 billion seed round at a $5.1 billion valuation for Ineffable Intelligence. The focus is reinforcement learning that requires far less human-generated training data than current approaches. Backers include Sequoia, Lightspeed, Nvidia, and Google. This is early-stage work with a long timeline to impact. But the directional bet matters: if reinforcement learning can produce capable agents without massive human-labeled datasets, the cost and speed of building specialized agents changes significantly. The skill that won't automate away regardless of how training evolves is the ability to set clear objectives, define what good looks like, and measure real outcomes.