Copilot Memory became generally available, with dreaming in public preview

On October 28, 2026, GitHub launched GitHub intelligence, a context layer for agents, made Copilot Memory generally available on paid plans and opened a public preview of dreaming.

Archive entry: this covers an event from October 28, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On October 28, 2026, GitHub launched GitHub intelligence, a context and learning layer for agents sourced from the graph of GitHub data. It combined memory, dreaming and code search that reached beyond the local checkout. The announcement is in the GitHub changelog entry on GitHub intelligence.

What changed

Copilot could remember personal preferences and repository conventions across sessions and clients.
Memories could be created in the Copilot app, Copilot CLI, VS Code, Copilot cloud agent, Copilot code review, Slack, Teams and GitHub Mobile.
Among Copilot CLI preview sessions that used memory, 93% showed better adherence to developer preferences and 42% showed more efficient workflows attributed to memories.
Dreaming reflected on past work to improve future agent sessions.
Dream types included personal memory curation, repository customization (which proposed updates to existing skills and custom instruction files) and code review customization.
Changes to checked-in skills and instructions were proposed as pull requests.

What it meant for people building with AI

Persistent memory cut down on re-explaining project conventions to agents every session. Because dreaming was opt-in and proposed its changes to checked-in skills and instructions as pull requests, those changes arrived as proposals rather than edits applied on their own.

Availability and controls

Copilot Memory was generally available on paid Copilot plans. Dreaming was in public preview. Personal memories could be viewed or deleted in Copilot settings.

Sources

GitHub intelligence puts your code, context, and history to work (GitHub, October 28, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.