Quickstart
Create an account, open your workspace, and get one real issue out of chat in about five minutes.
Playbooks
Ten workflows teams actually run, from a support escalation to a launch review, each a task you can start today.
Reference
Mode cheat sheet, terminology, every issue property, and the troubleshooting checklist.
What you get
Sessions with three modes
Every session runs in Ask, Plan, or Agent mode. Ask answers and changes nothing, Plan proposes before touching anything, Agent works the task end to end.
Issues and projects
An issue is the unit of work: one task, its properties, and the path from Triage to Done. Projects group issues into an initiative with a lead, dates, and a roadmap.
Documents with their own chat
Plans, specs, and decisions in a block editor. Each document carries a context chat anchored to its text, so questions get answered from the page, not from memory.
Connected apps and coding agents
Link GitHub, Linear, Notion, Slack, Google, and Cursor. Hand verified work to Cursor or GitHub Copilot, or let Bright’s own Engineer open the pull request.
Ways to use Bright
Ask, then Plan, then Agent
Research read-only, turn the answer into a reviewable plan, then approve it and let Bright create the project and ship the work.
One-shot
One Agent-mode prompt that covers research, planning, project creation, and implementation, with the artifacts saved for later.
Skills and named agents
Encode the review you run every release as a skill with an output contract, and give it to an agent with a point of view.
Keep exploring
- Where your code runs: engineering work happens in an ephemeral cloud sandbox, and nothing is durable until it is pushed.
- Context, permissions, and safe operation: what Bright can see, what a connected provider permits, and when a tool needs your approval.
- Models and Genius mode: when the more capable frontier model is worth its cost.
- Configure workspace settings: profile instructions, build defaults, members, labels, tags, and billing.