Agents are vertical. Organisations are horizontal

Longer work-rhythm context improves an agent's next-step prediction by 20%, but only when a worker switches between kinds of work.

The researchers organised two months of workplace telemetry into nested time scales, from single actions up to whole-day rhythms, and tested whether the larger patterns carried information the recent action stream did not. They did, but only at the seams: the added history helped when a person moved between kinds of work and added noise when the work simply repeated. That locates where the coordination, context and judgement Layer 1 names have to land, since an agent acting on the open file alone is least informed at exactly the transitions between work modes where horizontal work meets vertical agent capability.

Enterprise IT leaders find agent deployment is bottlenecked by organisation more than technology: agents attach to processes that cut across silos no single team owns.

Levie synthesises recurring themes from meetings with enterprise IT leaders now deploying agents, and the lead theme is that capturing the benefit is an organisational problem before a technical one: agents are most effective tied to a process, and processes run across silos, leaving unresolved who owns them and how they get deployed centrally. A second theme holds that token counts are the wrong measure and that outcome-based absorption has to be judged workflow by workflow, which resists top-down management. This is the first clean buyer-side account of the vertical-capability, horizontal-organisation mismatch Layer 1 names, sharpening the mechanism to agents attaching to cross-silo processes the siloed organisation never owned end to end.

The leading frontier model fully satisfies every requirement on only 3% of multi-week knowledge-work tasks.

The first independent, non-vendor evaluation of frontier models on realistic multi-week knowledge work scores them against a rubric built to catch output that looks polished but is wrong, and the current leader satisfies every requirement on just 3% of tasks. Vertical agent output still lands on a person to finish, check, and integrate before it is usable. That puts a measured ceiling on the absorption tax Layer 1 names: capability produced faster and more coherently than the horizontal organisation can review and integrate.

Citadel data: jobs in highly AI-exposed occupations are rising, not falling.

Software engineering postings up 18% YoY; customer service and accountancy also rising. Azhar reads the data through three mechanisms: complementarity, supervision overhead, demand expansion. Meaningful pushback on the simple substitution narrative; needs a Layer 1 reading that allows for compounding gain without proportional headcount compression.

Contested · evidence · 19 jun 2026Challenger, Gray & Christmas attribute roughly 40% of US May job cuts to AI, echoed in the 19 June Azhar roundup. The fresh layoff data cuts against the "rising, not falling" reading on the same AI-exposed occupations, so this card is held contested pending occupation-level corroboration.