MIG
Authenticate the agent and session, evaluate published policy and runtime context, then return Allow, Approval, or Deny before a protected tool executes.
- Agent-bound identity
- Shadow and Enforce modes
- Versioned decision evidence
House of Galatine / Applied AI systems
AI that does the work.
Control that keeps it accountable.
We build the infrastructure that governs consequential AI-agent actions—and the custom automation that helps businesses move faster.
Build useful systems◆Govern consequential actions◆Keep evidence attached◆Deploy around real work◆Build useful systems◆Govern consequential actions◆Keep evidence attached◆Deploy around real work◆
01 / What we build
Product and services
A focused product.
A practical delivery capability.
Automation
Turn repetitive or fragmented business work into AI-powered workflows shaped around your systems, constraints, review points, and measurable outcome.
- Workflow automation
- Custom AI agents
- Tool and business-system integrations
Measured, scoped, reviewable
We make the claim—and keep the boundary attached.
MIG’s internal Benchmark v1 recorded 1,600 measured HTTP requests with zero observed HTTP failures in the documented environment. Automation engagements begin with a defined workflow and measurable result.
1,600Measured benchmark requests
4–6Weeks for a scoped MIG pilot
01Workflow to prove value first
02 / Founder
Engineering from the consequence backward
Useful AI should be able to act. Responsible AI should be able to show who allowed it.
Indrooneel Panday
Mechanical engineer / Founder
House of Galatine brings product engineering and hands-on automation delivery together. We build around the operational outcome first, then make the system understandable, governable, and supportable.
- BEng Mechanical Engineering, Coventry University
- USPTO provisional patent pending
- Build3 Impact Accelerator
- Python / Neo4j / FastAPI / Modbus TCP
Choose the right starting point