The trust layer for AI agents
Multicorn builds the infrastructure that keeps humans in control of what AI agents can do on their behalf.
The company
Multicorn AI Pty Ltd is registered in Australia. We build open-source tools that give people visibility and control over the AI agents acting on their behalf.
Why Multicorn exists
AI agents are proliferating faster than the safety infrastructure around them. Today, agents can send emails, spend money, and access sensitive data with little oversight. Users deserve to know what these agents are doing and have meaningful control over it. Multicorn exists to close that gap. Not by slowing agents down, but by making their actions visible, auditable, and revocable.
An AI agent connected to your Gmail can read, send, and delete your emails with no oversight. An agent with access to your company's tools can spend money, publish content, and access sensitive data - all without asking. Shield gives you a consent screen to choose what each agent can do, a live dashboard to see what it's doing, and a kill switch to stop it instantly.
The founder
Multicorn was founded by Rachelle Rathbone. Rachelle is a backend engineer on Atlassian's Rovo Agents team, where she works on the infrastructure that powers AI agents at enterprise scale. Before moving into engineering, she spent years as a primary school teacher, which is why Multicorn's documentation is written for humans, error messages are actionable, and nothing requires a manual to understand.
That background shows in the product. The SDK is built by someone who understands agent infrastructure from the inside. The docs and error messages are written by someone who believes clarity is never optional. That experience is why Multicorn exists - she knew what good agent governance looked like, and saw that nobody had built it for everyone else.
Open source at the core
Multicorn follows an open-core model. The Shield SDK is free, MIT-licensed, and fully auditable. You can read every line of code, run the tests, and extend it for your needs. The hosted platform (dashboard, team management, compliance features) is how we sustain the business.
We believe trust tools should themselves be trustworthy. Open source is how we earn that.
What we stand for
Transparency
You should always know what an agent is doing, why, and what data it can access. No black boxes.
User control
People, not agents, decide what happens. Every action should be reviewable, every permission revocable.
Privacy by default
Agents should have the least access they need to do their job. Data stays where it belongs.
Developer experience
Safety tools only work if developers actually use them. We build for real workflows, not compliance theatre.