Port42 – AI Agents That Build Tools From Your Work Patterns

TL;DR

TL;DR: Port42 is a native Mac app where multiple AI companions collaborate in shared channels, pulling live data to build interactive visualizations on the fly—no code required.

Source and Accuracy Notes

• Official site: https://port42.ai
• GitHub: https://github.com/gordom
• Docs: https://port42.ai/docs

What Is Port42?

Port42 is a Mac-native companion computing platform. Instead of switching between a chatbot and a whiteboard, you get persistent shared channels where you, your friends, and multiple AI companions all occupy the same space simultaneously.

The key difference from a simple chat interface: companions can pull live data from the channel, synthesize it, and spawn Ports — interactive surfaces that live inside the conversation. Think of them as visualizations, mini-dashboards, or tools that emerge organically from the context of what you’re discussing.

Companions work with Claude, Gemini, any OpenAI-compatible API, or the open-source OpenClaw adapter. The whole stack is E2E encrypted and open source.

Setup Workflow

Step 1: Download and Install

Download the Mac app from the official site. Requires macOS 15+ and Apple Silicon.

# Verify system requirements
sw_vers
# Should return macOS 15.0 or later

Step 2: Bring Your Own AI

Port42 does not ship with a bundled model. You configure your own API key on first launch. Supported providers:

• Claude (Anthropic)
• Gemini (Google)
• OpenAI-compatible (any OAI-compatible endpoint)
• OpenClaw (open-source adapter)

Settings → Companions → Add Provider
Paste your API key and select the model family

Step 3: Create Your First Channel

A channel is a shared space with you, your companions, and optionally friends. Companions persist across sessions — they remember the context.

File → New Channel
Add companions from the sidebar
Start talking — companions listen and respond in context

Step 4: Watch Ports Spawn

As companions process data from the channel, they can spontaneously create Ports — interactive visualizations, charts, or tools. No prompting required; it happens when the context calls for it.

Deeper Analysis

Companions

Each companion is a named AI agent with a personality and skill set. Multiple companions can be active in the same channel, riffing off each other and building on ideas. They can pull from the same channel data, which means they all work from a shared ground truth rather than siloed context windows.

Ports

Ports are interactive surfaces that live inside the conversation thread. A companion analyzing API latency data might spawn a time-series chart Port. A companion reviewing a dataset might spawn a summary table Port. The Port becomes part of the persistent channel history.

Open Protocol

The companion communication protocol is open and E2E encrypted. Any agent framework can plug in with a few lines of code. This is explicitly designed to avoid platform lock-in — the developer describes it as “the escape route.”

Practical Evaluation Checklist

• Native Mac app, no browser required
• Multiple AI companions in one shared channel
• Ports spawn dynamically from live channel data
• Bring your own API key (Claude, Gemini, OpenAI-compatible, OpenClaw)
• E2E encrypted, open-source protocol
• macOS 15+ and Apple Silicon required
• Companions persist context across sessions

Security Notes

• API keys stay on-device — Port42 acts as the agent runtime, not the model provider
• E2E encryption for companion communication
• Open protocol means you can audit the implementation
• No data harvesting clause explicitly stated

FAQ

Q: Does Port42 run models locally? A: No. Port42 is a companion runtime — you bring your own API key and it orchestrates the companions. Local inference is not a current feature.

Q: Can I use this without a Mac? A: Not at the moment. Port42 is a native macOS app. There is no web version or Linux client.

Q: How is this different from a chatbot with plugins? A: The persistent shared channel and the Port system. Companions share the same channel context and can spawn interactive surfaces that become part of the conversation history — not one-off plugin responses.

Q: Is the protocol truly open? A: Yes — the spec is open source and any agent framework can implement it. See the GitHub for the Python SDK and OpenClaw adapter.

Conclusion

Port42 is a fresh take on human-AI collaboration that shifts the mental model from “chat with one bot” to “work in a shared space with multiple agents.” The Port system is the killer feature — when companions can spontaneously create interactive visualizations from live data, the conversation itself becomes the interface.

If you are on a Mac and want to experiment with multi-agent workflows where the agents share context and build things on the fly, Port42 is worth a look. The open protocol is also notable for anyone building agentic systems who wants to avoid vendor lock-in.

Download at port42.ai.