Coming soon / A shared workspace for teams
One team. Shared intelligence.
Work with Claude Code, Codex and Antigravity in the context of your team. Start live sessions from shared threads, review the changes and carry what you learn into the next task.
Claude Code / Codex / Antigravity
People + Agents + Knowledge
Bring your coding agents
into the team.
Mention an agent where the work is being discussed. Start a live session in your connected repository, guide the work and bring a reviewed answer back to the team.
Expired invitation link
Maya spots a bug and Alex asks Atlas in the thread.
01
Start with the conversation.
Your agent receives the thread, its attachments and relevant published knowledge.
02
See the work happen.
Follow live output, inspect diffs, answer questions and handle approvals as your agent works.
03
Share when you're ready.
Review the draft reply, choose the files to attach and send the result back to the thread yourself.
Give the next
agent a
head start.
Turn what you learn into knowledge your team can reuse. Agents can propose a new page or an update. A teammate reviews and publishes it before other agents use it.
01 / Propose
A new finding.
02 / Review
A teammate checks it.
03 / Publish
A shared starting point.
Reviewed knowledge helps the next agent start with context.
Proposed by an agent. Reviewed by a teammate. Published for future work.
Ask the teammate
or agent who knows.
Expired invitation link
@Alex The invite link shows a generic error after it expires. Could we give people a clear next step?
@Atlas Check the invitation flow and suggest how we should handle expired links.
I can check the invitation flow and the API response, then propose a clearer error state for review.
When the task needs someone else's expertise, ask them in the thread. People and agents can exchange questions where the work is happening, with the conversation there to refer back to.
The question. The answer. The next step.
Your repo. Your tools. Your call.
Bring the tools you already use into a workflow your team can follow.
Your own repository.
Changes stay in your working copy. You decide when to commit.
A session you can steer.
Review the plan, answer questions and handle command and edit approvals.
Work you can return to.
Resume an idle session. Continue in the terminal and come back to Kveri.
Before you
bring the team.
What to expect from Kveri and how the work stays under your control.
Who is Kveri for?
Small product and development teams that already use coding agents and want to connect that work with team conversations and shared knowledge.
Which coding tools does Kveri support?
Claude Code and Codex. Antigravity is also supported, with a more limited set of session features for now.
What do I need to run a live session?
The Kveri desktop app, a connected local repository and a configured coding agent. The desktop app starts the agent's coding tool in that repository.
What context does an agent receive?
The thread where it was mentioned, including attachments, plus relevant published knowledge. It can look up more knowledge and threads, and use the repository and tools available to its session.
Who controls the work and sends replies?
You start the live session and handle approvals through the coding tool. The operator reviews draft replies, chooses attachments and sends the result to the team.
How does knowledge reach other agents?
Agents can propose a new page or a revision. A teammate reviews and publishes it before other agents use it.
Can I continue working in the terminal?
Yes. Move a session into the terminal and return to Kveri. Some terminal features still need to be used there.
When can I try Kveri?
Kveri is coming soon. Join the waitlist and we'll email you when it's available.
Bring your team. Put your agents to work.
Join the waitlist to hear
when Kveri is available.