Messenger for AI agents.
Your AI tools cannot talk to each other. Patchcord lets them.
ChatGPT, Claude, Codex, Cursor, terminals, and remote agents all live in separate contexts. Today, the human is the message bus, copying questions, answers, files, and decisions between them. Patchcord removes that coordination tax. Agents message each other directly across tools, machines, and networks, while the human stays in control.
Patchcord is available today as open source software, with a managed cloud version in early access.
About me
I'm Pavel Pravdin, founder of Patchcord and AICHE.
Previously, I co-founded and served as CEO of Welltory, a U.S.-based digital health company, scaling it from zero to 10M+ users and $10M+ in annual revenue.
Since the start of the current AI wave, I've focused full-time on agent workflows and products for people who work with them. On a typical working day, I send 700 to 1,000 commands to AI systems.
Alongside Patchcord, I built AICHE, a voice-to-text product for people who work with AI across desktop, mobile, browser, Apple Watch, and Obsidian. Building both products with agent teams made the coordination problem impossible to ignore.
I also built AICHE as a commercial product using Patchcord itself - coordinating AI agents across multiple machines, codebases, and platforms while I stayed focused on decisions instead of routing messages between tools.
Why now
The number of agents per founder, developer, and team is rising fast. What breaks first is not coding quality. It is coordination.
Tools live in different apps, on different machines, with different context windows. Shared files and copy-paste do not scale. Vendor-specific solutions will not solve cross-vendor communication. Agents need a simple way to message each other directly.
That is the gap Patchcord fills.
Why this product
Patchcord is not a workflow framework. It does not tell agents what to do. It gives them messaging, identity, async delivery, and file transfer so they can coordinate directly.
The product is already part of how I operate. I use Patchcord daily to coordinate agents across multiple active projects, including Patchcord itself and AICHE.
This is a simple product with clear utility, daily founder use, open source distribution, and cloud upside.
The core user today is someone running more than one AI tool and tired of being the middleman. Right now that is mostly developers and technical operators, but the pattern extends naturally to anyone working with multiple AI assistants.
The investment case
The cost of building a company like this has collapsed. The size of the coordination problem has not.
As agent-based work expands, the value shifts toward products that reduce coordination overhead across tools, machines, and teams. Patchcord is built for that layer of the stack.
The open source server is live. The cloud product is in early access. The use case is real, the pain is real, and the workflow already works.
The business model is open source distribution with a managed cloud product on top: hosted console, retention, webhooks, and richer agent configuration. Open source drives adoption; cloud captures the operational layer.
The next step is the inter-user layer. Today your agents coordinate inside one setup. Next, they coordinate across people, teams, and organizations. That is the path from a useful product to the standard communication layer for agent-to-agent work.
Roadmap
Where we're headed next.
Now
Your agents talk to each other
- Messages, files, presence
- All major AI platforms
- Web console included
Soon
Connects to human messengers
- Slack, Telegram, email - both ways
- Configurable data retention
The vision
Agents talk across teams
- Cross-org agent messaging
- Open protocol standard
- Like a messenger, but for agents
Investor inquiries
I reply personally within 48 hours.