Tavi MCP
Tavi, inside your AI assistant.
Connect Tavi to Claude, ChatGPT or Codex, then source candidates, research them and build shortlists in the same conversation where you already work.
Ask in plain language. Tavi does the legwork.
- 01
Connect.
Add Tavi as a connector in your AI assistant and sign in with your Tavi account.
Tavi Connector
Connected to your Tavi workspace. What should we look into?
- 02
Ask.
Describe the role you’re hiring for, or share a job post, the way you’d brief a colleague.
You
Find senior ML engineers in Europe who ship. Here’s the job post.
- 03
Tavi gets to work.
It searches in the background and comes back with candidates ranked on your brief, each with a short research note.
Tavi Connector
Here are the first 12 candidates, ranked on your brief, each with a short note on why they fit.
- 04
Refine.
Rank on your own criteria, drop what doesn’t fit and dig into the profiles that stand out.
You
Re-rank on startup experience and drop the agencies.
- 05
Keep it.
Save shortlists to your Tavi workspace. Your work stays there and updates as people move, instead of disappearing when the chat ends.
You
Save the top ten to my Research Engineer list.
Research Engineer10 candidates · Shared with your team in Slack
Updates as people move
Just ask.
See a market
“Who are the platform engineers at fintech scale-ups in London?”
Run a deep search
“Find a founding ML engineer for this brief.”
Rank on your criteria
“Re-rank on startup experience and drop the agencies.”
Find warm paths
“Who on my team knows anyone on this list?”
Build shortlists
“Save the top ten to my Research Engineer list.”
Pick up where you left off
“Open last week’s map and show me what’s changed.”
One workspace, every surface.
The MCP server and the Slack agent share one workspace. A shortlist you build in Claude is there for your team in Slack, and the other way round. Each company gets its own isolated environment, so your data never touches another customer’s.