About
Liftli is built by one person, in public.
Liftli is an AI content strategist for LinkedIn, X and Substack. It is not a scheduling tool and it is not a chat wrapper. It runs inside the AI assistant you already use — Claude, Claude Code, ChatGPT, Codex or Cursor — over the Model Context Protocol, so it can see the work you are already doing and turn it into posts that sound like you wrote them, because you did.
Who builds it
Liftli is built and run by Oded Tsamir, working solo. It is not a venture-backed company and there is no team behind the logo — it is one operator, shipping in public, using the product every day on his own feed.
The reason it exists is narrow and personal. Oded spent years shipping product for other people and almost none of it was visible: the experience, the opinions and the receipts were real, and the LinkedIn profile said nothing. Every tool he tried wanted him to sit at a desk and describe his week to a text box — but the week was already over, and the ideas had already happened, in calls, in commits, in things said out loud walking to the car. So he built the thing that catches them.
What that means for you
Being small has consequences worth stating plainly. Support is a real person answering email, usually within a day, not a ticket queue. The roadmap moves in days rather than quarters. There is no enterprise procurement process and no sales team — you can start free with no card, and the paid plans are self-serve. It also means Liftli is early: some surfaces are in beta, and that is labelled where it is true.
How Liftli handles your work
Liftli reads a source when you point it at one, through a connector you already control. Nothing is scanned in the background, and nothing is published without you approving it. Full detail is in the privacy policy and the terms.
For agents and developers
The publicly callable surface is documented at /openapi.json,
the agent-facing summary is at /ai-agents.md, and setup
instructions live on the developer page. The MCP endpoint is
https://mcp.liftli.ai/mcp.