AI Partners.Send a brief
← All field notes
04 · AGENTS

From assistants to agent teams: the fluency you'll need next.

Field note6 min read

If the prompt age was about one person typing into one chatbox, the next phase is about orchestrating several AI agents that run for hours and sometimes act on your behalf. Anthropic's 2026 Agentic Coding Trends report describes the shift bluntly: from single assistants to coordinated agent teams, with the human moving from doing the work to designing and supervising the systems that do it.

This does not retire the 4Ds. It raises the stakes on every one of them.

Delegation gets harder, not easier

Anthropic's own research found a striking gap: developers use AI in roughly 60% of their work, but report being able to fully delegate only 0–20% of tasks. The judgement of what to hand over — and what must stay human — becomes the core skill, not an afterthought. As one engineer put it, they use AI mainly where they already know what good looks like, an ability built by doing the work "the hard way" first.

Discernment becomes oversight at scale

Judging one output is one thing. One engineer described their future as "taking accountability for the work of 1, 5, or 100 Claudes." That is discernment scaled into a supervision discipline — evaluating quality across many parallel agents you didn't watch write the code.

The goal isn't to remove humans from the loop. It's to make the loop scale.

Diligence becomes architecture

When agents act with autonomy, accountability has to be designed in from the start — security, boundaries, review gates — not bolted on after. The report's organisational priorities for 2026 read like diligence at the system level: coordinate multiple agents, scale human oversight, embed security from the earliest stage.

Where this leaves you

The individual who only prompts is not just behind today; they are unprepared for a world of agent teams. The team without shared fluency will struggle to supervise one agent, let alone a hundred. Building the 4Ds now is not a nice-to-have for 2026 — it is the prerequisite for using any of this safely.

The most useful step is also the simplest: find out where you actually stand. Then build deliberately from there.

Content informed by Anthropic's AI Fluency resources. Further reading: Anthropic — 2026 Agentic Coding Trends Report · Anthropic — How AI Is Transforming Work

© AI Partners 2026 · Senior on every job · PrivacyAI that actually ships.