Speed up software development
Use GitHub Copilot in the editor, the command line and on GitHub.com, including the cloud agent that prepares pull requests, while human review, tests and pipelines stay in charge of what ships.
The problem
Developers spend much of their time on boilerplate, tests, upgrades, reading unfamiliar code and small fixes that wait in a backlog. Review queues and pipeline failures slow delivery further.
AI coding tools can help, but organizations need to decide which plan, which surfaces and which policies apply, and how to keep generated code under the same review and quality gates as human code.1
The approach
- In the editor: GitHub Copilot offers inline suggestions and chat. In Visual Studio Code, an agent works in the same workspace as the editor, terminal, tests and debugger, with a Plan option for planning, a model picker, bring-your-own-key support and MCP servers for external systems.
- On GitHub.com: the Copilot cloud agent (formerly called the coding agent) researches a repository, plans and implements changes in an ephemeral cloud environment, runs tests and linters, and opens a pull request. GitHub states a 59-minute session limit, one branch and one pull request per task, and that it works only with repositories hosted on GitHub. It is not included in Copilot Free.
- Review: Copilot code review summarizes pull requests and suggests improvements; people still approve.
- Plans (checked 26 Oct 2026): GitHub lists Copilot Business at $19 and Copilot Enterprise at $39 per granted seat per month, with monthly GitHub AI Credits allowances; code completions are not billed in credits on paid plans.
- Work tracking and pipelines: teams on Azure DevOps keep Boards, Repos and Pipelines; the Azure DevOps MCP Server lets AI assistants query work items and pipelines.
When a simpler route is enough: inline completions and chat in the editor cover many teams; the cloud agent pays off when there is a backlog of well-described, testable tasks.12345
Conceptual architecture
Diagram as a list
Agents & automation
- Copilot cloud agentImplements tasks in the background and opens pull requestsConnects to Repositories on GitHub (or Azure Repos)
AI models & platform
- GitHub Copilot (suggestions, chat, agent)Generates and explains code and testsConnects to Repositories on GitHub (or Azure Repos)
Developer tools
- Visual Studio Code and Visual StudioWhere developers write, run and debug codeConnects to GitHub Copilot (suggestions, chat, agent)
- Copilot CLIAgent help from the terminalConnects to Repositories on GitHub (or Azure Repos)
- Repositories on GitHub (or Azure Repos)Source of truth for code
- Pull request review (human plus Copilot code review)Gate before mergeConnects to CI/CD pipelines
- CI/CD pipelinesBuild, test and deploy
- MCP servers (for example Azure DevOps MCP Server)Give agents access to work items and other systemsConnects to GitHub Copilot (suggestions, chat, agent)
Security, identity & governance
- Organization policies for CopilotDecide which features and models are allowed
Technologies and their roles
GitHub Copilot1
AI coding assistant and agent
Suggestions, chat, cloud agent, code review and CLI across plans.
Visual Studio Code and Visual Studio 20262
Developer environment
VS Code hosts the Copilot agent with planning, model choice, BYOK and MCP servers.
Azure DevOps5
Boards, Repos and Pipelines
Free for up to five users with basic features; an MCP server exposes work items and pipelines to AI assistants.
What you need first4
- A Copilot plan for each developer, chosen by the features needed (the cloud agent is not in Copilot Free)
- Organization policies set for allowed features and models
- Repositories on GitHub for the cloud agent
- Automated tests and linters so agent changes are checked
- Branch protection and required human review before merge
Risks and how to reduce them
- Generated code merged without proper review1
- Require pull request review and passing checks; GitHub says humans stay responsible for approving agent work.
- Agent tasks time out or sprawl3
- Write small, testable tasks; a cloud agent session has a 59-minute limit and one pull request per task.
- Credit spend above plan allowances4
- Monitor GitHub AI Credits; GitHub lists overage at $0.01 per credit for Business and Enterprise pools.
- Repositories outside GitHub cannot use the cloud agent3
- Use editor-based Copilot for those repositories or plan a migration.
Documented examples
Woven by Toyota, Inc.
PilotWoven by Toyota: AI agents that fix MISRA code violations
81.5% (measured)
Automotive, Japan
Related
Sources
Thank you. Your correction was sent.
The editors check it against the sources. If you left an email address, they may reply about it.