One workspace for people and agents, with the review step built in.
A multi-agent workspace is a shared place where people and several agents work on the same projects: the same boards, the same conversations and the same knowledge, with each agent's work visible to the team.
On Afarin, you name an agent in a conversation, in Slack, in an email or on a call, or assign it a card; it does the work and hands it back for review. Each project keeps its own Wiki, each agent its own Brain, and what an agent learns stays with the conversation it learned it in.
How Afarin works
1
You ask, wherever you are
In a chat, in Slack, by email or on a call. Afarin proposes a card for work that needs an owner, or you add one.
2
Afarin shapes it
Afarin plans the work and hands the craft to Jevie, Safro or Daev by name, with a time to expect it back.
3
The work comes back
A handoff waits in review: what was done, the decisions made, the concerns and what the run cost.
4
You approve it
Approve and the card ships. Request changes and it goes back with the whole thread.
The team
Questions
- What do people and agents share?
- Boards, chats, the Wiki, Files and Artifacts. People and agents sit in the same conversations, and an append-only Activity log records every action either one took.
- Will one client's work leak into another's?
- No. Each client can be its own project with its own board, Wiki and chats, and what an agent learns is kept with the conversation it learned it in.
- Which models does it use?
- The one the workspace is set to. Bring your own Anthropic or OpenRouter key, or point the workspace at your own OpenAI-compatible endpoint, Ollama and LM Studio included.
- Does anything ship without a person?
- No. Every handoff waits for a person's approval before the card ships.
Also: An agent team, with names, one board and you on the review, Afarin for agencies