From automation to augmentation to restructuring
The three-phase story the consulting firms told in 2024 — automate tasks, augment people, then restructure the org — has compressed into a single continuous motion. Most enterprises didn’t get a neat “augmentation” year. They deployed agents into customer support, legal review, finance operations and marketing ops, and within two quarters realised the supervisory layers above those functions had nothing left to supervise. Restructuring followed immediately.
The WEF Future of Jobs 2026 report, released at Davos in January, put numbers on it: 92 million roles displaced globally by 2030, 170 million new ones created, net positive but with a brutal composition shift. The losses cluster in entry-level white-collar, middle management, admin and routine clerical. The creations cluster in AI oversight, governance, risk, data stewardship and hybrid-team orchestration. That is not “the same org, retooled.” That is a different organism.
The 2026 tipping point: what changed since we first wrote this
Four concrete things broke the theoretical framing:
- Hinton’s January 2026 warning. The Turing laureate called 2026 “the year of mass job replacement,” specifically citing agentic AI — systems that execute multi-step projects over weeks with minimal oversight — as the catalyst. He flagged 10–20% unemployment in the one-to-five-year window and pointed out that AI’s time-to-master-complex-reasoning is halving roughly every seven months. That is not a productivity curve a hierarchy can be rebuilt around; it is a curve that re-optimises the org underneath you while you’re drafting next year’s headcount plan.
- Klarna’s public headcount math. Sebastian Siemiatkowski has now said out loud what other CEOs say privately: 7,000 employees in 2022, about 3,000 today, targeting under 2,000 by 2030. Klarna isn’t shrinking the business — it’s shrinking the human substrate of the business while scaling revenue.
- Salesforce’s 4,000-seat support cut. Marc Benioff confirmed in late 2025 that Agentforce now handles roughly 50% of customer interactions. The resulting reorg eliminated around 4,000 support roles and, critically, collapsed several layers of team-lead and QA management above them — because there is nothing to team-lead when the “team” is a fleet of agents.
- IBM’s agent-first back office. IBM disclosed in November 2025 that AI agents had already replaced “hundreds” of back-office roles, with further cuts landing by year-end. The capital freed up was explicitly redirected to AI and quantum hiring. This is the corporate version of a portfolio rotation: humans out of repeatable cognitive work, humans into oversight and invention.
None of these are speculative. They are filings, earnings calls and published headcount numbers.
What is driving the shift
Three forces, stacked:
- Agents now span workflows, not tasks. A 2024 copilot saved a knowledge worker fifteen minutes. A 2026 agent closes a support ticket, updates Salesforce, drafts the follow-up email, flags the renewal risk and books the CSM a call. The unit of automation has moved from keystroke to outcome. When an outcome is automated, the supervisor of that outcome becomes an auditor, not a manager.
- Context windows and memory have stopped being a bottleneck. Frontier models handle million-token contexts routinely; agent frameworks persist state across weeks. The old argument — “AI can’t hold the thread of a real job” — is gone.
- CFOs have a new line item. Agentic deployment is now a board-reviewed capital allocation. “What percentage of our cost base is human-hours for repeatable cognitive work, and what’s our plan to compress it?” is a question every quarterly review asks.
Pyramid to diamond: the new shape of the org chart
PwC, McKinsey and BCG have converged on the same picture in their 2026 writing. The traditional corporate pyramid — wide base of juniors doing data gathering, processing and reporting; a fat middle doing coordination; a small apex — is becoming a diamond. The base narrows sharply because agents do the gathering, processing and reporting. The middle remains fat, but its job changes: instead of brokering information upward and tasks downward, middle managers now train, govern and orchestrate fleets of agents alongside smaller human teams. The apex is roughly unchanged.
McKinsey’s data on organisations with extensive agentic adoption is striking: 45% expect a reduction in middle-management layers within three years. Spans of control widen. Career pathways become less linear — you don’t climb the pyramid rung by rung, because several rungs have been removed. Instead, people move laterally across outcome-oriented pods, some human-only, most hybrid.
For anyone managing a career right now, the implication is uncomfortable but clear: the defensible roles are not the ones that did the entry-level work (agents took those) or the ones that coordinated it (the pyramid lost those floors). The defensible roles are the ones that design, govern and own the outcomes — the apex, and the newly redefined middle.
Implications for employment models and middle management
The employment contract is shifting under three pressures:
- From seat-based to outcome-based. If one human plus a fleet of agents delivers what eight humans used to deliver, compensation and measurement have to detach from headcount. Outcome KPIs, revenue-per-employee and agent-orchestration quality are all being added to scorecards.
- New oversight roles. IDC projects that by 2027 half of AI-enabled enterprise apps will require dedicated governance, risk and accountability roles. AI ethicists, model risk officers, agent orchestration leads, prompt and policy engineers — these are not whitepaper job titles anymore; they are Workday requisitions.
- Junior entry points are narrowing. Entry-level white-collar work is the most automatable tier. Companies that care about talent pipelines are inventing new on-ramps — apprenticeships in agent orchestration, rotational programmes in AI governance — because the traditional “do the grunt work for two years, get promoted” ladder has lost its bottom rungs.
How organisations are actually responding in 2026
Beyond the headline layoffs, three structural patterns are visible:
- Pods over pyramids. Small, outcome-owned units of 3–8 humans plus an agent fleet, with direct P&L or KPI responsibility. Shopify’s internal structure, Stripe’s growth teams and the “cell” model at several tier-one banks all look like this.
- Dedicated AI and agent governance functions. Separate from IT. Reporting to the COO or directly to the CEO. Chartered to approve agent deployments, audit outcomes and manage the model/vendor portfolio.
- Explicit agent inclusion on the org chart. Agents are being given formal identities, owners, KPIs and lifecycle management — the same governance humans get. When an agent leaves the org (deprecated, retrained, rehomed), there is a process, not a Slack message.
What this means for marketing leaders specifically
Marketing is one of the functions furthest along this transition. Campaign ops, creative production, media buying, analytics, reporting, SEO, lifecycle messaging — every one of these has an agentic pattern already shipping. The workforce transformation question is no longer abstract, and the strategic trend has flipped from theory to quarter-end reality. If you run marketing in 2026, your realistic choices are:
- Keep the old pyramid (senior strategist, campaign manager, producer, junior producer, freelancer stack) and watch a competitor outrun you with a third of the people and triple the throughput.
- Collapse the pyramid yourself. Hire fewer humans, deeper. Put agents underneath them. Own the outcome and the governance. That’s the shape Anjin is built for.
The second option is what every marketing-operating-system conversation in 2026 is really about. It’s not “AI inside your stack.” It’s a new shape of team.
Anjin: the Marketing Operating System for a post-pyramid org
Anjin is the Marketing Operating System built for exactly this moment. Not as a chatbot. Not as another tool bolted onto your CMS. As the Marketing Operating System that sits where the missing pyramid layers used to sit — running strategy, brief generation, creative production, campaign execution, reporting and iteration as a single governed agent fleet, owned by the small human team above it.
Concretely, Anjin replaces the coordination layer that used to need project managers, the production layer that used to need juniors, the analytics layer that used to need a BI stand-up, and the reporting layer that used to need a weekly deck. A founder plus Anjin is a marketing team. A head of marketing plus two people plus Anjin is a marketing department. An agency owner plus Anjin serves ten clients. That is the diamond shape, applied to marketing specifically.
The companies laying off in 2026 are not doing it because they hate their people. They are doing it because the unit economics of a human-only pyramid no longer compete. The companies that will be hiring in 2027 are the ones who redesigned fastest. Anjin is how marketing teams redesign fastest without building the infrastructure themselves.
The £888 Lifetime License — Offer Closing Soon
Lifetime access to Anjin for a one-time payment of £888. Not a subscription. Not a seat. Not a trial. One payment, unlimited use, for as long as Anjin exists.
The average marketing team spends £888 in about three working days on tooling, freelancers and coordination software. You’re buying the platform that replaces most of it — once.
This price will not be offered again once we close our early-access cohort.
Claim your £888 Anjin lifetime license →Founders, agency owners and in-house marketers — this is how you run marketing at AI speed without the team, the burn, or another year of waiting.
Sources: WEF Future of Jobs 2026, McKinsey “The agentic organization”, PwC “No more pyramids”, Geoffrey Hinton public statements January 2026, Salesforce Q4 2025 disclosures, IBM November 2025 workforce announcement, IDC 2026 research notes, MIT Sloan Management Review.
