Matrix/Organization

Organization

How organizations adapt to the age of agents. From "buy licenses" to "agent fleet management".

4capabilities20levels60practices60guides
The matrix · full map
Capability ↓
Maturity →
L1 · Stage 01
Assisted
L2 · Stage 02
Delegated
L3 · Stage 03
Systematic
L4 · Stage 04
Governed
Sweet spot
L5 · Stage 05
Self-improving
01·15 guides
AI Adoption Model
How your organization rolls out AI tools - from individual experiments to org-wide strategy
Licenses bought, little changes
3 practices·3 guides
Pilot teams and a champion
3 practices·3 guides
A platform team owns the tooling
3 practices·3 guides
AI-first is the culture, not a memo
3 practices·3 guides
The org is built around agent throughput
3 practices·3 guides
02·15 guides
Knowledge Management
How institutional knowledge is captured, shared, and made available to both humans and agents
Knowledge lives in people's heads
3 practices·3 guides
Docs and ADRs get written
3 practices·3 guides
Documentation is infrastructure
3 practices·3 guides
Context flows to agents automatically
3 practices·3 guides
The knowledge base updates itself
3 practices·3 guides
03·15 guides
Team Structure & Roles
How teams are organized and what roles exist to support AI-augmented engineering
Classic roles; seniors mop up AI code
3 practices·3 guides
Champions and the first context engineers
3 practices·3 guides
Harness and platform engineering are real jobs
3 practices·3 guides
Developers manage fleets, not files
3 practices·3 guides
Agentic engineers orchestrate; anyone contributes
3 practices·3 guides
04·15 guides
Tech Debt & Modernization
How AI accelerates paying down tech debt and modernizing legacy systems
Debt piles up, untouched
3 practices·3 guides
Debt is at least triaged
3 practices·3 guides
Agents pay debt down in the background
3 practices·3 guides
Dead projects modernize for pennies
3 practices·3 guides
Debt near zero, patched 24/7
3 practices·3 guides
Climb the matrix

You don't have to figure this out alone.

Every level in this matrix has a path. Read the playbooks the teams that have climbed it wrote. Run the assessment with our consultants. Start where you are.

Live with Visdom

Book an AI Maturity Assessment session with your team.

We walk you through all four perspectives, score where you actually are, and leave you with a 90-day plan to climb in the dimensions that matter most.

Book an assessment See what's included90-day plan - scored assessment - coaching
Author Commentary

August produced the best-documented answer yet to "what does the org chart look like".

Zalando's snapshot is the case study of the month: a proxy running since January 2024, ~2,000 monthly active users across 250+ teams, cost tracking and prompt-caching handled by the platform so developers get the savings without doing anything - and deliberately no mandated tool. Enablement is a real programme with countable attendance: a weekly guild, guided hackathons, hands-on labs. Ramp's Inspect is the other shape: 75% of merged PRs authored by an internal agent maintained by five and a half people, with 200+ agents built on top by other teams. Both numbers are self-reported by interested parties. The transferable part is not the percentage, it is the architecture: a central place that hands out identity, cost tracking, sandboxing, observability and evals by default, so that declaring an agent takes minutes instead of weeks.

Rachel Laycock gave the role its name. In The Conductor Developer the bottleneck is human *attention*, not coding capacity, and the skills that matter - attention management, energy management, deciding under incomplete information - are precisely the ones no career ladder currently measures. Her follow-up, Citizens Build, Agents Execute, Experts Govern, is about where value moves rather than job titles: experts shift from features to guardrails, platforms and practices. And Rahul Garg supplied the hard constraint on all of it in The Orchestrator's Tax: what degrades a multi-agent setup is context pollution, not token cost. "Tokens are spent once. Context shapes every decision that follows." Cap batches at two to four, stop polling for status, treat overlapping file ownership as a signal to consolidate.

The anti-pattern of the month deserves naming, because it is spreading. Metering AI adoption by token spend or seat activity is Goodhart-complete: within weeks you get engineers burning tokens for the metric, people put on performance plans for not using a tool while delivering fine work, and others buying personal subscriptions to route around a corporate mandate. Zalando's counter-example works precisely because it inverts this - the tool choice is free and the outcomes are measured - and even there, engineers gamed the risk classifier by splitting PRs. Their own conclusion is the one to write on the wall: AI amplifies the good and the bad practices you already have. It does not install new ones. Worth pairing with the labour data, which refuses to be simple: AI led all stated US layoff reasons for a fifth straight month while total announced cuts hit a two-year low, and roughly a third of roles eliminated primarily because of AI were later refilled.

Other perspectives