AI enablement · Software teams

One AI-assisted engineering workflow your team agrees on, measured on real work

If your developers use different AI coding tools in different ways and nobody can say whether it helps, you can commission an engineering enablement programme. I baseline how work flows today, pilot one shared AI-assisted workflow with one team on a real repository, measure cycle time, review effort and defects against that baseline, and leave a playbook your team maintains.

This is a good fit if…

  • Your engineers use a mix of coding assistants, chat tools and agents, each with their own habits, and reviewers cannot tell what was generated or how it was checked.
  • You have bought AI coding licences and want to know whether they change delivery, not just how many people log in.
  • Pull requests have grown larger and review has become the bottleneck since AI tools arrived.
  • You want an agreed team practice for prompts, review, tests and ownership before rolling AI tools out to more teams.

Look elsewhere if…

  • You want a slide-deck training day with no change to how work is done. This programme changes one real workflow.
  • Your developers need to learn to build and run LLM features, not to use coding tools. Use build internal AI engineering capability instead.
  • You need an AI-generated application made production-ready. Use take an AI-generated application into production instead.

What you get

Engineering enablement pilot and rollout

  • A baseline of how work moves today: cycle time, pull-request size, review time and rounds, change-failure and escaped defects.
  • One AI-assisted workflow agreed by the pilot team and used on a real repository, with explicit review, test and ownership rules.
  • Mechanical guardrails in CI (tests, types, linters, required checks) so quality does not depend on reviewer vigilance.
  • A measured comparison of the pilot against the baseline, including what did not improve.
  • A team playbook and a rollout decision for other teams, owned by an internal lead.

How it runs

  1. 01

    Baseline

    I interview engineers, read recent pull requests and pull delivery data from your repositories and tracker, so we know how work flows before anything changes.

  2. 02

    Design the workflow with one team

    With one pilot team, we choose the tools, the task types AI is used for, the review and test rules and how AI-assisted work is labelled.

  3. 03

    Pilot on real work

    The team uses the workflow on its normal backlog for at least two full sprint cycles. I pair, review and adjust the rules weekly from what goes wrong.

  4. 04

    Measure, decide and hand over

    We compare the pilot with the baseline using the same measures, write the playbook, and you decide whether to roll out, revise or stop.

What needs to be in place

  • An executive sponsor and an internal owner, often a staff engineer or developer productivity lead, who will own the playbook.
  • One pilot team with a real backlog and a repository with CI.
  • Approved AI coding tools, or a decision on which ones the pilot may use, with licences and data terms in place.
  • Read access to repository and tracker data needed for the baseline.

Not included

  • Guaranteed productivity gains or a promised payback period. The programme reports measured change, including none.
  • Ranking individual engineers by AI usage or output.
  • Hiring, restructuring or performance management decisions.
  • Tool procurement or vendor negotiation.

Which engineering enablement route fits?

Start here whenSignature output
Engineering enablement (this page) Several tools, inconsistent habits, no measurementBaseline, one-team pilot, playbook and rollout decision
Claude Code enablement You have chosen Claude Code and need controlled repository useRepository pilot, tool-permission checklist, review protocol, rollback exercise
GitHub Copilot enablement You have Copilot seats and need them to improve work without review debtRepository pilot, review-effort baseline, usage playbook
Coding-agent rollout Agents are starting to open pull requests on their ownTask scope, permission model, test requirements, stop conditions
AI-assisted code review Generated code is overloading review and releasesReview workflow, seeded-defect evaluation, release controls
Internal AI engineering capability Engineers must learn to build and own LLM featuresSkills-to-work matrix, paired backlog, graduation checklist

The gap this programme closes

Most engineering teams are somewhere between “we use AI tools” and “we know what AI changes about how we ship”. One engineer uses a coding agent for whole features. Another uses chat for boilerplate and pastes the results in. A third has stopped using anything after a bad experience. Pull requests have got bigger. Reviewers cannot tell which code was generated or how it was checked. Leadership has licence numbers and anecdotes, and no way to say whether delivery improved.

This programme replaces that with one workflow the team agrees on, piloted on real work and measured against a baseline. It is not a training day and it does not depend on one product.

What changes

The programme works on four levers, in this order:

Coding workflows. Which task types the team uses AI for and which it does not. How work is broken down so generated changes stay reviewable. How prompts, project instructions and context files are kept in the repository and reviewed like code. How AI-assisted pull requests are labelled.

Mechanical guardrails. Quality should not depend on a reviewer noticing a subtle bug in a 900-line generated diff. The pilot puts checks where they cannot be skipped: tests on changed code, types, linters, required CI checks and pull-request size limits. This is the method I have published as Vibes Inside Guardrails: let people move fast inside constraints that machines enforce.

Engineering culture. A clear rule that the person who opens a pull request owns it, whoever or whatever wrote it. Agreed boundaries for what agents may do without approval. Review that focuses on design and risk, because the guardrails catch the mechanical errors.

Measurement. A small set of measures taken the same way before and during the pilot, so the result is evidence rather than enthusiasm.

How the programme runs

Baseline. I read recent pull requests, talk to engineers and pull data from your repositories and tracker: cycle time from first commit to production, pull-request size, time to first review, review rounds, change-failure rate and escaped defects. This is the comparison point.

One-team pilot. With one team, we design the workflow and the guardrails, then use them on the team’s normal backlog for at least two full sprint cycles. I pair with engineers, review alongside them and adjust the rules every week from what goes wrong.

Measure and decide. We take the same measures again and compare. The report says what improved, what did not and why. You decide to roll out, revise or stop.

Playbook and rollout. The pilot team writes the playbook with me. Rolling out to further teams is planned as a separate step led by your internal owner, using that playbook.

The signature deliverable

You end with an engineering enablement pilot and rollout: the baseline, the measured pilot result, the playbook and a rollout plan. Illustrative example of the measurement summary:

MeasureBaselinePilotNote
Median cycle time (first commit to production)measuredmeasuredSame repository, same task mix
Median pull-request size (lines changed)measuredmeasuredSize limit introduced in pilot
Review rounds per pull requestmeasuredmeasured
Escaped defects per sprintmeasuredmeasuredTracked for both periods
Share of pull requests labelled AI-assistedn/ameasuredAdoption, not success

Illustrative example. The table shows the format; values come from your own data.

Ownership and handover

The playbook, the CI checks and the measurement queries belong to your internal owner from the start. The pilot team presents the result to the next team, not me. When I step back, your owner can run the measures, change the guardrails and onboard a new team.

Choosing a narrower route

If you have already chosen a platform, the Claude Code and GitHub Copilot pages go into that product’s settings and controls. If agents are opening pull requests on their own, start with coding-agent rollout and team practices. If review load is the main pain, see AI-assisted code review and release controls. If your engineers need to learn to build and run LLM features, see build internal AI engineering capability.

Questions buyers ask

How long does the programme take?

It depends on team count and the workflows in scope, and the written scope states it before you commit. The structure is always the same: baseline, a one-team pilot running for at least two full sprint cycles, then measurement and a rollout decision. Further teams are a separate, smaller step once the playbook exists.

Why measure cycle time and defects instead of tool usage?

Usage tells you people opened the tool, not whether work improved. Lines generated rewards the wrong thing. Cycle time, review effort and escaped defects tell you whether changes reach production faster without more breakage, and they are measurable from your own repository and tracker before and after.

Do you train on a specific tool?

The programme is tool-neutral. It works with whichever coding assistants or agents you have approved. If you have already standardised on Claude Code or GitHub Copilot, the platform-specific pages go deeper on that product's settings, permissions and review features.

Who does the work?

I do, personally: the baseline, the workflow design, pairing with the pilot team and the measurement. If a specialist would help, for example on your CI platform, they are named and approved by you before they start.

What if the pilot shows no improvement?

Then you have learned that cheaply, on one team, with data. The report shows which measures moved and which did not, and why. Sometimes the right change is in tests, CI speed or pull-request size rather than more AI. Stopping is a valid outcome.

Improve one engineering workflow

A short, non-confidential description is enough to start. I read every brief personally and reply within two business days, including when the answer is that I am not the right fit.

Step 1 of 2 · The basics