What Is Orlo?
Orlo is a governed domain AI control plane built for a specific operational surface:
the path between model output and real-world consequence
That path shows up in concrete places:
- a fraud or support decision served through a live inference endpoint
- a retrieval step pulling from documents that change over time
- an agent asking to call a write-capable tool
- a production correction that should become a better dataset instead of disappearing into a ticket queue
Orlo exists to make that path measurable, governable, and improvable.
Orlo answers that with a combination of:
- evaluation
- deterministic validation
- deployment controls
- retrieval and attribution
- feedback-driven improvement
- agent-step governance
Orlo Open Core
Open Core exposes reusable building blocks:
@orlo/shared@orlo/validation@orlo/runtime-adapters@orlo/agent-sdk@orlo/studio
These are useful to developers, ML engineers, and platform teams building on Orlo concepts.
Orlo Platform
Orlo Platform is the full managed platform:
- task and dataset management
- evaluation and recommendation
- deployment and inference
- retrieval and document ingestion
- feedback promotion
- agent governance APIs
This is the surface for teams that want the full evaluation, deployment, and governance loop in one system.
What Orlo Is Not
- not a general-purpose agent runtime
- not a workflow orchestrator
- not just an LLM gateway
- not only a prompt management tool
- not an endpoint security product
Orlo is strongest in the governed middle layer between model providers, production systems, and consequential domain use cases.