Cross-AI context management

Your AI tools finally remember your work

Relay quietly captures decisions, tasks, and context

scattered across ChatGPT, Claude, Gemini, and your coding agents

then hands that memory to every fresh conversation.

Relay - Stop repeating yourself to every AI | Product HuntRelay — AI Memory & Context Sync

Chrome extension · MCP for your IDE · Free to start

Works with
OpenAIChatGPT
ClaudeClaude
GeminiGemini
GrokGrok
PerplexityPerplexity
DeepSeekDeepSeek
OpenAIChatGPT
ClaudeClaude
GeminiGemini
GrokGrok
PerplexityPerplexity
DeepSeekDeepSeek
OpenAIChatGPT
ClaudeClaude
GeminiGemini
GrokGrok
PerplexityPerplexity
DeepSeekDeepSeek
AntigravityClaude Code
CursorCursor
CodexCodex
AntigravityAntigravity
WindsurfWindsurf
GithubCopilotGitHub Copilot
AntigravityClaude Code
CursorCursor
CodexCodex
AntigravityAntigravity
WindsurfWindsurf
GithubCopilotGitHub Copilot
AntigravityClaude Code
CursorCursor
CodexCodex
AntigravityAntigravity
WindsurfWindsurf
GithubCopilotGitHub Copilot

Saving to Relay

My App Project

OpenAIchatgpt.com
Browser chat

Let's continue building the auth flow. We decided to use Supabase and the user table needs...

I'll help with the auth flow. Based on the Supabase setup, we should first define the user table schema with RLS policies...

Message ChatGPT...
AntigravityClaude Code
IDE agent (via MCP)

get_brief("My App Project")

✓ Loaded project brief

Decisions: Supabase for auth, PostgreSQL...

Tasks: Implement auth flow, add RLS policies...

Constraints: No third-party auth providers

Based on the brief, implementing RLS:

CREATE POLICY "users_own_data"

ON public.users

USING (auth.uid() = id);

The problem

Context doesn't survive the switch

Every new chat starts from zero

You re-explain your project, your stack, and your constraints to the same AI — again.

Your decisions live in five apps

What you chose and why is buried across ChatGPT threads, Claude chats, and your editor.

Your AI forgets after every session

The context you built up yesterday is gone today — for you and for your coding agents.

Works withChatGPTClaudeGeminiGrokPerplexityDeepSeekClaude CodeCursorCodexCopilotAntigravity
Coming soonTelegramWhatsAppDiscordGmailCalendarNotionLinearJiraSlackGitHub

How it works

A calmer loop than copy-pasting transcripts

01

Relay captures as you chat

Work in any supported AI. Relay quietly extracts decisions, tasks, and constraints from the conversation — nothing manual.

02

Context flows both ways

Your project brief updates itself across sessions and tools. What you decide in the browser reaches your IDE agent via MCP — and vice versa.

03

One click fills the next chat

Open a fresh conversation. Insert your project brief. Keep building exactly where you left off.

Browser chats

Relay

Captures context from chats and syncs it to your agents

IDE agents

What Relay does

Context that moves with you

AUTO-CAPTURE

Quietly saves what matters from every AI chat

Work in ChatGPT, Claude, or Gemini. Relay captures decisions, tasks, and constraints automatically — no manual saving.

PROJECT BRIEFS

One-click context restoration in fresh chats

Your project brief updates itself as you work. Open a new chat and inject the full context instantly.

MCP INTEGRATION

Your coding agent reads and writes project memory

Claude Code, Cursor, and any MCP-compatible agent connect directly. They share the same brief as your browser chats.

CROSS-SURFACE SYNC

Decisions flow between tools automatically

A choice made in ChatGPT surfaces in Cursor. A constraint set in Claude Code stays in sync with your next browser session.

MCP Integration

Your coding agent,
fully in the loop

MCP (Model Context Protocol) is an open standard that lets AI agents read and write structured memory. Relay implements it as a bridge between your browser sessions and any IDE agent.

Quick install
$ npx @onrelay/wizard
1

Browser chat → Relay captures context

2

Relay → project brief updates automatically

3

IDE agent → reads latest brief via MCP

writes back decisions and progress

Supported agents

Claude CodeCursorCodexWindsurfOpenCodeGemini CLIVS Code (Copilot)TraeAntigravityZedContinueAiderAmpVoidClineRoo CodeAugmentKilo CodeCopilot (CLI)Amazon Q

Pricing

Simple, honest pricing

Start free. Upgrade when Relay is part of how you work — no jargon, cancel anytime.

MonthlyYearly -17%

Free

$0/ forever

Try Relay free — no card needed

  • Try Relay memory with your AI chats
  • 2 projects to keep your work organized
  • One-click capture from ChatGPT, Claude & more
  • MCP for Claude Code, Cursor & coding agents
  • Enough reads & writes to feel the difference
Get Started
Most Popular

Starter

$7/ month-17%

Billed $80 / year

For people who use AI tools every day

  • Never re-explain a project to your AI again
  • 5 projects with long-term memory
  • Context updates itself as you work — automatically
  • Richer briefs that carry continuity between sessions
  • Enough usage for daily AI work, every day
Get Starter

Pro

$10/ month-17%

Billed $120 / year

For heavy users who want the most automation

  • Everything in Starter, with the most headroom
  • 15 projects, highest limits everywhere
  • Highest-quality AI model for summaries and briefs
  • Most proactive updates with conflict resolution
  • Relay agent without thinking about quotas
Get Pro

* Pricing subject to change during beta. Early users keep their rate.

FAQ

Questions & answers

Your AI tools finally
remember your work

Relay quietly captures decisions, tasks, and context scattered across ChatGPT, Claude, Gemini, and your coding agents then hands that memory to every fresh conversation.

Free forever · No credit card · Takes 60 seconds