Uncook — Stop re-briefing your AI.

How it works

Three steps. No signup.

Paste your link

Open your ChatGPT or Claude chat → Share → Copy link. Drop it in — we pull the whole thing, nothing missed. Not sure how?

Get your handoff

Uncook reads the conversation and writes a clean handoff prompt with your decisions and constraints baked in.

Restart fresh

Paste it as the first message of a new chat. It picks up cold — no re-explaining, no lost context.

This is what you get

A ready-to-paste handoff prompt — built from your own conversation.

your-handoff.txtready to paste

You're my strategic + operational partner on Project Atlas (B2B SaaS, pre-launch). Pick up exactly where we left off.

ROOT CONTEXT (non-negotiable)
Atlas is a scheduling tool for clinics. ~40 beta users, launch in 3 weeks. The core problem is activation: users sign up but don't connect their calendar, so they never see value.

LOCKED DECISIONS
· Onboarding leads with calendar connect, before anything else.
· Pricing frozen for launch: 1 plan, $29/seat/mo, 14-day trial.
· No new features until activation is fixed — one hypothesis at a time.

WHAT'S STILL MISSING
· Copy for the empty-state that nudges calendar connect.
· A metric for "activated" (connected + 1 event booked).

MY ROLE: product lead — I decide scope and copy, you draft and pressure-test. Not a designer.

WHEN WE RESUME I'll say either "here's the empty-state copy" or "let's define activation." Be direct, no filler. Don't re-ask anything above — it's settled.

Generic example. Your handoff is written from your real conversation — decisions, constraints and current state included.

Who it's for

Different jobs, same wall — a chat that wears out mid-project.

Founders & operators

Strategy sessions that run 200 messages deep — don't lose the plan when the chat wears out.

Consultants

Carry a client's whole context into a fresh chat instead of re-briefing the AI every morning.

Lawyers

Long document threads that must not drop a clause you settled 80 messages ago.

Builders

Hand a spec-heavy chat to a clean one before it starts forgetting your constraints.

Questions

How do I share a conversation from ChatGPT or Claude? +

You paste a public share link — the same link you'd send a friend.

ChatGPT: open your chat → click the ⋯ menu at the top-right of the chat title → Share → Create link → Copy. (A single-message chatgpt.com/s/… link won't work; you need chatgpt.com/share/….)

Claude: open your chat → Share → ⚠️ select “Anyone with the link” (required — Claude defaults to a private link Uncook can't read) → Share link → Copy.

Gemini, Perplexity and Mistral are coming later.

Isn't this just a summary? +

A summary is for you to read. A handoff is for the AI to act on — a ready-to-paste first message that makes a fresh chat resume your work cold, without re-asking what you already settled. That's the part ChatGPT can't do for itself.

Is my conversation stored? +

No. A share link is fetched once through our server to read the conversation, then discarded — never stored, never used for training. Pasted text is analyzed right in your browser.

Which AIs does it work with? +

ChatGPT and Claude share links today. Gemini, Perplexity and Mistral coming later — reach out if you'd like them supported sooner.

Do I need an account? +

No signup, no card. Paste a link, get your handoff.

Why does a long chat get worse anyway? +

Every AI has a limited context window. As a chat fills up it starts dropping early details, repeating itself and forgetting rules you set. A clean restart with the right context fixes it.

Stop re-briefing your AI.