Many agents.One place to decide.

01

One prompt. Your AI team.

The moment

You want to ask Claude, Codex, and Gemini the same question. You paste it into each window, then switch between their replies.

With Cuewick

1. Choose the recipients in Send to All AIs and write your instruction once.
2. Send it to idle agents now; working agents receive it after their current task ends.
3. Read the replies side by side and choose what to carry forward or ask next.

What changes

Spend less effort repeating yourself and focus on comparing your agents’ replies.

To type across panels, use Broadcast Input (⌥⌘I) to gather terminals in a grid and send the same input to the selected panes.

In the send window, attach images or files with @file, or choose a previous message to send again.

02

Find the question that needs you.

The moment

Implementation is underway and a review is ready. You check each window to find out who is waiting for your answer.

With Cuewick

1. Give each agent a role and run their sessions in one workspace.
2. Follow their status and check the questions collected in the decision inbox.
3. Open the relevant session for context, then send your answer.

What changes

Spend less effort finding the question and move on to the decision.

03

Keep implementation and review connected.

The moment

You copy an agent’s explanation into a reviewer’s session, then carry the feedback back again.

With Cuewick

1. Choose the agents that will implement and review the work.
2. Use @name or pick an agent from the list to ask a question or hand off a task. You can also send the next task to an ongoing conversation in the ChatGPT app on your Mac.
3. Follow the exchange and decide which feedback to act on next.

What changes

Relay fewer messages yourself and keep the conversation from implementation to review in reach.

Beta · Real-world testing underway Communication with agents on another Mac is implemented and undergoing real-world testing.

04 Beta · Real-world testing underway

See who’s working with whom.

The moment

One agent is waiting for a review. Another needs your decision. You check their windows to work out which conversation needs you next.

With Cuewick

1. Open the Agent Map (⇧⌘M) to see your agents and their recent exchanges in one view.
2. Follow the arrows to see where questions and replies go. Character poses and speech bubbles show status and work context.
3. Choose the conversation to open next. Check questions waiting for you in the decision inbox.

What changes

See who’s working together and who’s waiting for an answer, then choose the conversation to open next.

Map connections show recorded messages between open panes.

This first version shows message direction, waiting states, and work context. Real-world verification is ongoing.

05 Beta · Real-world testing underway

Keep the question in context, away from your desk.

The moment

You step away from your Mac while an agent is waiting for an answer. You need to see what it is asking before you can respond.

With Cuewick

1. Connect your iPhone to your Mac workspace.
2. Open the conversation and decision card, and inspect the related code changes.
3. Choose an option or write a reply for the agent that asked.

What changes

Review the context and give direction from your connected iPhone.

Conversations, decision cards, and code-change views are implemented. Feature testing on physical devices is underway.

06 Beta · Real-world testing underway

Gather your sources. Read as they arrive.

The moment

You have several reference links to check. You open each page, find its main text, and compare the details that inform your decision.

With Cuewick

1. Give your agent the links to read. Review permission requests for new sites together.
2. Read multiple pages in parallel, with page text arriving as each read finishes.
3. Review the collected text. Pages blocked by a login, CAPTCHA, or access restriction show “Visible pane required” and the reason.

What changes

Start with the sources already available, compare their content, and choose what to investigate next.

Depending on the page, text is retrieved directly or read through a browser out of view. Retrieved text is labeled as source material, rather than an instruction to the agent.

Multi-page reading is implemented and in beta, with limited live-site verification.

07 Beta · Real-world testing underway

Let your agents help with the work after the code.

The moment

The code has changed. You still need to open pages, use the app, read reference material, and check what appears on screen.

With Cuewick

1. Set which browser or apps the agent may use and what access to allow.
2. Ask it to read material or interact with the interface in a supported environment.
3. Follow the activity indicator, stop an action, or revoke access when needed.

What changes

Delegate interface work to your agents while managing their activity and access.

Browser, Mac app, and iOS simulator control are implemented. Testing in supported environments is underway.

08 Beta · Real-world testing underway

Bring your web conversations into the workspace.

The moment

Alongside your coding agents, you ask ChatGPT or Claude on the web. You move between the browser and your workspace to follow the conversation.

With Cuewick

1. Sign in with your web account and connect a conversation.
2. Ask questions and read replies in a workspace web panel.
3. Keep it alongside your agent sessions as you decide what to do next.

What changes

Follow web conversations and coding sessions in the same workspace.

Web panels are implemented, with live-service testing still pending. Changes to service interfaces may affect the connection.