Maturity Matrix
Matrix/Delivery Management

Delivery Management

How we manage delivery in the age of agents. From human PR review to autonomous delivery pipeline.

4capabilities20levels61practices61guides
The matrix · full map
Capability ↓
Maturity →
L1 · Stage 01
Ad-hoc
L2 · Stage 02
Guided
L3 · Stage 03
Systematic
L4 · Stage 04
Optimized
Sweet spot
L5 · Stage 05
Autonomous
01·15 guides
CI/CD Pipeline
Speed and reliability of your build-test-feedback loop for AI-generated code
CI runs; everyone waits
3 practices·3 guides
Ten minutes, basic caching
3 practices·3 guides
Five minutes, incremental by default
3 practices·3 guides
Two minutes in an isolated microVM
3 practices·3 guides
Feedback in seconds, capacity on demand
3 practices·3 guides
02·15 guides
Merge & Deploy
How PRs flow from creation to production - throughput, automation, and conflict handling
Every merge has a human in it
3 practices·3 guides
A queue does the rebasing
3 practices·3 guides
Policy decides what merges, and when
3 practices·3 guides
Green auto-merges straight to production
3 practices·3 guides
A thousand merges a week, agent-driven
3 practices·3 guides
03·16 guides
Metrics
What you measure to understand AI-assisted engineering productivity and quality
DORA at best; AI still unmeasured
3 practices·3 guides
Token spend is finally on a dashboard
3 practices·3 guides
You know your cost per merged PR
3 practices·3 guides
Auto-approve rate is a managed number
5 practices·5 guides
Cost per feature, value per token
2 practices·2 guides
04·15 guides
Governance & Compliance
Controls around AI-generated code - licensing, security scanning, and audit trails
Personal subscriptions, no policy
3 practices·3 guides
A policy exists and spend has caps
3 practices·3 guides
Every agent action leaves a trail
3 practices·3 guides
Provenance is cryptographic, checks automated
3 practices·3 guides
Compliance watches the regulators for you
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 2026 update: July moved the delivery bottleneck to a place DORA never measured - review capacity.

RIT's study of 25,264 agentic PRs found 78.9% pass through a single reviewer, and adding reviewers changed nothing; GitLab's research says AI raises PR volume but not cycle time - the queue just moved to the review/security/deploy gates. The month's starkest datapoint: curl suspended all vulnerability-report intake for five weeks because AI slop pushed its confirmed-vulnerability rate from ~15% to under 5%. This is what "generation is free, verification is not" looks like operationally. The answer the market converged on is back pressure by design: merge queues for agent fleets, mergeability + post-merge outcomes instead of benchmark scores (OpenAI formally retracted SWE-Bench Pro), and provenance that can say which agent did what - now extending from cryptographic traces down to commit-to-prompt lineage and decision ledgers.

Governance had its most consequential month yet, mostly via defaults. Claude Code flipped its default permission mode to Manual (July 3); GitHub shipped enterprise managed settings across every Copilot surface with the exactly right framing: any client outside the policy is a gap. Two corrections to keep your compliance calendar honest: the EU AI Act's August 2026 high-risk milestone largely evaporated (Digital Omnibus finalized June 29 - deferred to Dec 2027 / Aug 2028), while China's agent-specific rules took effect July 15 as the world's first dedicated agent regime. And the sovereignty file got its epilogue: Fable 5 came back July 1, but nobody un-learned the lesson - API stability itself is now a procurement line item (DeepSeek retired its legacy endpoints with 4 days' notice and announced peak-hour surge pricing). Stripe Minions remains the L5 north star; the new homework is proving your review capacity can absorb the fleet you are about to unleash.

Other perspectives