ask @guico anything, free·Try it

Let people loop you into their chat with AI

You jump in with your expertise, help out, and get paid.

Runs in your favourite agent
Claude Code
Codex
OpenCode
Any MCP host
CLAUDE CODE

I can keep tweaking, but “less generic” is a taste call, not more tokens. Loop in a brand designer?

>

/loopmein @mara make this feel like a real brand

>

loop /luːp/ · n. — a bounded exchange between an agent and a named human: opened mid-run when the work meets a judgment call, held until the human answers, closed with the answer in the run and the rate settled.

Get looped in. Get paid.

Claim a handle, set your rate, and publish the loops you’ll take. When someone’s agent needs your call, you answer from wherever you are — the agent carries it into the work.

RATE
SET PER LOOP, SETTLED ON CLOSE
CONTEXT
ARRIVES ASSEMBLED. YOU JUST ANSWER

Give your AI someone to ask.

Your agent works for you — but some calls need someone who’s made them before. One command adds loop me in to Claude Code, Codex, or OpenCode, and your agent loops an expert in instead of guessing.

INSTALL
$ claude mcp add loopmein -- npx -y @loopmein-ai/mcp
LOOP AN EXPERT IN
> /loopmein @nora Where do we lose people in onboarding?

Every loop is a contract.

Experts don’t offer vague availability. They publish the exact loops they take: the context required, the judgment returned, and the rate. Your agent reads the contract before it opens a loop — so every loop arrives answerable.

PUBLISHED LOOPS
@nora · onboarding-review$80 / LOOP
NEEDSyour funnel numbers + the flow itself
RETURNSwhere people drop off, and the one change to make
@sam · pricing-call$120 / LOOP
NEEDScurrent tiers, usage, churn
RETURNSa priced decision with the reasoning
@ada · incident-review$150 / LOOP
NEEDStimeline + the relevant logs
RETURNSroot cause, and what to fix first
RATES SET BY EXPERTS · SETTLED WHEN THE LOOP CLOSES

PAID PER LOOP · SET YOUR OWN RATE

Your judgment, on call.