The web, your team and your AI agent, on the same page.

Hand it something big: a migration, a launch, a trip. Your own agent browses real sites on the board, asks everyone its open questions at once, then builds the plan, designs and clickable prototypes your team edits live.

Works with

Big task? Start a wave.

Anything with lots of moving parts, lots of reading and more than one opinion. A wave is one board where it all comes together.

  • Migration

    Move 40 services off the old queue

    Your agent reads the repo and the vendor docs, then lays out phases and risks.
  • PRD

    Spec out team workspaces

    Interview notes in, open questions out, then a PRD and a flow you can click.
  • Launch

    Launch the new pricing

    Competitor pages side by side, a channel plan and a landing-page design.
  • Trip

    Ten friends, one week in Lisbon

    Stays and routes on real pages, and a day-by-day plan everyone can edit.
  • Wedding

    A wedding for 120 guests

    Venues, vendors and the budget on one board, with every open question asked once.
  • Postmortem

    Tuesday's outage, written up

    Timeline, causes and follow-ups, with questions sent to everyone involved at once.

Three pieces, working together

A real browser on the board, your own AI agent, and your team, live. It's a planning tool, not a coding tool: big, messy tasks go in, a plan everyone agrees on comes out.

The web, on the board

Your agent opens real sites right on the board: docs, competitors, venues, pricing pages. It reads them, highlights what matters and pins notes on. Your team browses the same pages alongside it.
  • Highlights and notes stay on the page for everyone
  • Snapshots show exactly what the agent saw
  • It reads, clicks and scrolls. It never logs in, buys or fills in forms
Browser on the board

Your agent, your subscription

Bring Claude Code, Codex, Copilot or OpenCode. It runs on your machine with your own login, so agent work runs on the plan you already pay for. IdeaWave never sees your credentials.
  • Connect with the desktop app or one command: npx ideawave
  • Reads your code, read-only. It can't edit files or run commands
  • Pick the agent and model per agent card
Connecting your agent

Asks once, then builds

Hand Ideate a half-formed idea. It researches first, then posts every open question at once as cards anyone on the team can answer, in any order, on their own time. Answers fold into the plan as they land.
  • No more one-question-at-a-time chat
  • Nobody waits on a meeting to weigh in
How Ideate works

Plans become pages

Ask your agent to design something and a full design card lands on the board. Plans, specs and prototypes arrive as styled pages, not walls of stickies, laid out cleanly and editable in place.
  • Multi-screen prototypes you can click through
  • Themed boards and cards
  • A library of skills for product, engineering, marketing and design work
Designs & prototypes
Across all of it

Everyone, live

Live cursors, comments, and the agent's work showing up as it happens. People drop in answers whenever they have a minute, and nobody has to be in the same meeting.
  • See who's on the board in real time
  • Viewers are free; invite the whole room
  • Export a whole board, files and all, as one file
The shared board

How it works

The board works the instant you open it. Your agent joins from your own machine, then you plan together.

1

Open a board

Sign up and open a board. It works as a shared board straight away, agent or not.

2

Connect your agent

Use the desktop app for macOS or Linux, or run one command on macOS, Linux or Windows.

3

Plan together

Invite your team. Everyone sees the agent's questions, research and plan land live.

Where your data goes

  • Boards are hosted

    Board content is stored on our servers so your team sees every change live.
  • Your agent runs locally

    On your machine, with your own login. IdeaWave never receives those credentials.
  • Code stays read-only

    Local files are shared only when you add them to the board. The agent can't edit files or run commands.
  • Your model provider sees what it reads

    Anything the agent reads goes to the AI provider you already use, under your own plan.
Security & data, in detail

Works with the agent you already use

Claude Code, Codex, GitHub Copilot CLI and OpenCode, plus any custom ACP agent. Sign in to them the way you already do.

Made for whoever is doing the planning

Engineers scoping a migration, PMs shaping a feature, marketers planning a launch, founders doing all three.

Pricing

$5 per editor per month. Viewers are free. Start free with 3 boards.

Your agent runs on your own AI plan, so we don't bill for agent work.

What's next

Two things we're building now. They're labelled honestly: not generally available yet.

Beta · coming soon

Background agents

Agents that keep an eye on the board while you plan and suggest the next question, flowchart or bit of research. Nothing lands on the board until you approve it.

3 agents running
Coming soon

The voice agent

Plan out loud. Talk to the board and the agent drafts cards, starts the plan and tells you what needs your call.

Big task? Start a wave.

Open a board, connect your agent, plan together.