Your team. Your agents.
One place to ship.

Bring your people, coding agents and project context together.
Plan the work, run it in parallel and review what ships.

Make the first run feel effortless. Use the customer feedback, simplify setup, and check the recovery paths.
Working for ·
A working replica of Alan using public sample content. A parent session splits an onboarding redesign across Claude Code, Codex, Cursor, OpenCode, and Command Code sessions that run and finish while you watch. Choose a session to open its transcript, switch between your sessions and your team’s, and open the workspace panel for the artifact or terminal.

Works with the agents and tools you already use

  • Claude Code
  • Codex
  • Cursor
  • opencode
  • GitHub
  • Slack

01 / Hand off

Give it the goal.
Not the next prompt.

Start with a task, the relevant context and a clear outcome. Ask Alan to split the work into parallel sessions, each with the agent and environment you choose.

Different agents. One shared brief.

Pick Claude Code, Codex or Cursor for each session. The brief, the diffs and the history stay on one task.

YouAlan
Studio
MineTeam
Today
Simplify the onboarding flow
Check the first-run edge cases
Review the interface changes
Write the release notes
Sessions / Check the first-run edge cases
Check the new onboarding flow. Cover an empty workspace, a returning user, and a failed connection.
Codex
Worked with the project context
Read the onboarding changesRead Welcome.tsx · WorkspaceSetup.tsx
Added recovery coverageEdit onboarding.test.tsx
Ran the focused checkspnpm test -- onboarding

The checks cover the first visit and the recovery paths. A failed connection keeps the setup in place and gives the user a way to retry.

First-run checks ready to reviewonboarding.test.tsx
onboarding.test.tsxView changes

02 / Review

See the change.
Check the evidence.

Ask agents to run tests and browser checks, then review the diff, report and recordings together. Keep the merge in your normal pull request review.

First visitPassed
acme.dev/onboarding
Your workspace is ready
Connection retryPassed
acme.dev/onboarding/connect
Connect your codeGitHubConnectedConnected to acme/web
Returning userPassed
acme.dev/
Your workspace is ready

3 agents · 3 flows · cloud sandboxes3/3 checks passed

Verification reportRequested by you · 18:41

  • First visitRecording 0:04 · screenshotPassed
  • Connection retryRecording 0:06 · screenshotPassed
  • Returning userRecording 0:05 · screenshotPassed

Pull requestReview required

Simplify the first-run experience #42

onboarding-polish → main · 2 files · +34 −3

  • Typecheck
  • Unit tests · 18 passed
  • Browser checks · 3 flows

Illustrative run on a sample app · checks run when you ask

03 / Away

Close the laptop.
Keep cloud work moving.

Cloud sessions run independently of your computer. Check the conversation, answer a question or review the result from your phone when you're away.

Codex · 00:20

Keep the old setup route behind a flag while the new flow rolls out?

Keep it behind a flagRemove it
Reply…
07:50 · answered from phone

Cloud runs on Pro and above · illustrative timeline

04 / Shared context

Keep the reasoning
with the work.

Give agents access to your codebase, specs and earlier decisions. With code search and memory enabled, they can find relevant work and trace it back to its source.

Code index and memory on Pro and above

GitHub, agent runs, docs, Linear and Slack feed Alan's code, product context and memory layers, and Alan brings that context to Claude Code, Codex, Cursor and OpenCode.

Context, with a sourceIllustrative task · API rate limiting

Team decision

Use a token bucket for burst traffic.

The billing service already handles bursts with an atomic Redis script. Reuse that approach for the public API.

Source: Billing retry limits · Payments team

Ready for the next session

  • The task brief and acceptance criteria
  • Relevant code and earlier changes
  • Past decisions, with references

Give the next agent the reasoning behind the work.

05 / Team

Fast agents.
Faster teams.

Bring a teammate into the session. Read the same conversation, steer the agents together and keep the task, decisions and results in reach of the next person.

Ship the onboarding redesignShared with Onboarding4 in session
  1. MayaLet's get the onboarding redesign out this week. UI and edge cases can run in parallel.
  2. Alan split this into 2 sessionsClaude Code · UI changesCodex · Edge cases
  3. Claude CodeUpdating the welcome screen and the setup steps.Editing Welcome.tsx
  4. CodexCovering an empty workspace, a returning user and a failed connection.Running onboarding.test.tsx
  5. DevI'll review the copy as it lands.
  6. CodexKeep the old setup route behind a flag while this rolls out?Needs a decision
  7. MayaYes, behind a flag.
  8. Simplify the first-run experience #42+34 −3 · checks passed · review required
  9. Dev approved #42 through pull request review

Illustrative session · shared sessions on Team and above

Explore Alan for Teams

06 / Setup

Your computer or the cloud.
Your agents either way.

Use an installed agent in your own checkout, or choose a cloud sandbox for work that needs its own machine. Pick the setup that fits the task.

Where it runsYour computerFree · $0The cloudIndividual · $10 a month
Runs onThe Alan desktop appCloud sandboxes
Your codeRuns in your checkoutCloned into your sandbox
AgentsThe ones you already pay forYours, or Alan-hosted models
Computer asleepWork waits for your computerWork can continue
Project contextSession history; code search and memory on paid plansCode search and memory

Team from $10 per seat for shared work · Enterprise on contract

07 / FAQ

Before you
hand it over.

How is this different from running my agent on its own?
Alan brings sessions, tasks, project context and reviewable results into one workspace. Run agents in parallel, invite a teammate into a shared session and use cloud sandboxes for work that needs to run independently of your computer.
Which agents and models can I use?
Alan supports Claude Code, Codex, Cursor, opencode, Gemini, Copilot and Alan's own agent. Each session can use a different agent. Availability depends on your plan and which agents are installed or configured. Paid plans also include Alan-hosted model usage.
Does it run tests on every change?
No. Agents run your tests and browser checks when you ask, or as a step in a workflow you set up. Reports land on the task, next to the diff.
Where does my code go?
Local sessions run on your computer, in your checkout, using the agent you configure. Cloud sessions run in sandboxes on paid plans. Your configured agent and tools may send code or context to their providers. Session history stays in Alan, and merges go through your normal pull request review.
How does Alan use past team work?
With code search and memory enabled, agents can retrieve relevant code and earlier decisions with source references. Retrieval respects access permissions. Connect your project context and enable the relevant features for your team; a new session can then build on what was already learned.
What happens when a cloud sandbox is paused?
Sleep can preserve files and running processes when supported by the sandbox provider. Shutdown keeps files but ends running processes. Rebuilding a sandbox is different from resuming it, so review the action before continuing.
How do I start?
Sign up free, download the desktop app and start a local session with an available agent. Individual adds cloud runs, code search and memory; Team adds shared sessions and collaboration, from $10 per seat a month on Pro. You can also book a demo to walk through a repository with us.

Set the direction.
Let the agents run.

One workspace for your people, your agents and the work ahead.

Free to start · Paid seats from $10 a month