ServicesCoding agent tools

GitHub Copilot within the policies you already have

We set up Copilot and its cloud agent to fit your organisation's policies, repository guidance and review process, then measure a pilot before it goes further.

Talk to an engineer

Example: what a repository holds for Copilot

.github/copilot-instructions.mdBuild and test steps, code style

.github/instructions/api.instructions.mdApplies only inside src/api

.github/workflows/copilot-setup-steps.ymlSets up the cloud agent's environment

.github/CODEOWNERSNames who reviews the agent's pull requests

Required checks and branch protection are left untouched.

What a first implementation project delivers

We start with one repository and a few pieces of work typical for the team, such as fixing a small bug that a failing test already shows. Before the pilot begins, you and we agree what success means and who makes the call.

At the end you have a working setup on that repository and a written decision on whether to expand, to which repositories and on what conditions.

Build and test guidance the cloud agent can follow, kept in the repository
Approval rules for the agent's pull requests, agreed with your reviewers
A record of accepted changes, review effort and cost for each task
The configuration and a runbook your team can use without us

Your GitHub settings come first

Copilot works within the rules set in your GitHub organisation. That is why we read through those rules before we touch anything: the organisation's settings, the policies on each repository, branch protection and CI.

Policies that turn Copilot features on or off, organisation-wide and per repository
Custom instructions in every repository on how its code is built, tested and written
An agreed list of the repositories and data the work covers

When the cloud agent opens a pull request

The cloud agent's pull requests are treated like every other change: a person reviews each one, and the required checks have to pass. We agree with your reviewers what to pay attention to, and when a change should go back to the agent.

The CI checks you require remain in force
Every reviewer can see what the agent was asked to do
What reviewers find, plus rework and reverts, broken down by team, task and model

A pilot measured against your baseline

Before the pilot starts, we measure a baseline. The pilot then runs on work that is typical for your teams, and its results are compared with those starting figures. Reports cover teams, tasks and models. No engineer is rated individually.

How developers learn to work with Copilot

Developers pick up Copilot fastest by working on issues that are actually open in your backlog. They learn to choose between chat, edits and the agent for a given task. We also show how to write up an issue clearly enough for the cloud agent to finish it. Reviewing the pull requests it opens is covered as well. Training runs remotely or on site in Germany or Poland, in German, English or Polish.

What your organisation keeps

The Copilot policies and the repository custom instructions
Runbooks for adding further repositories and teams
Measurements from the pilot, taken on your teams' work

What the numbers need to show

The number of the cloud agent's pull requests that are merged with no rework
How long a review takes when the agent opened the pull request, and when a person did
What seats and usage cost, set against the work that is merged

Questions

Does Copilot suit the way we work and the tools we use?

We check this before anything else. We go through your organisation settings and branch protection, your CI, and whatever tools connect to it. If Copilot turns out to be a poor fit for some repositories or types of task, the pilot makes that visible before a broader rollout.

Some of our teams work with other coding agents. Does that matter?

No. Work that spans several tools belongs to another of our services: Coding agent adoption.

Do you mean GitHub Copilot or Microsoft 365 Copilot?

GitHub Copilot, the assistant and agents software teams use for code. Microsoft 365 Copilot is another product, built for the Office applications, and it is not something we implement.

Related services

Coding agent adoption

Agents taken from a few pilot teams to the rest of engineering, with training and one shared setup.

Harness engineering

The setup that tells an agent how to build, test and change your code, kept in version control.

Evaluations on your repositories

Tools, models or settings tested side by side on changes you have already merged, before you commit to one.

From the blog

GitHub Copilot pilot metrics

What GitHub's Copilot usage metrics measure, where they stop, and how to combine them with your own review and rework data to decide whether to continue.

Branch protection for agent pull requests

The GitHub branch protection and ruleset settings that should apply when coding agents open pull requests, and how to check that they actually hold.

Talk to an engineer

Bring one piece of work you would like to improve. The call takes 30 minutes and needs no access to your systems or your data.

Book a 30-minute call

Booking runs on Google Calendar.

Prefer to write first? Leave your email and a few lines, and we reply with times.