Platform capabilities

Orlo helps a team decide how AI should work, use that decision in production, and improve it from real results.

What you want to do	How Orlo helps	Start here
Choose a model for a task	Compare candidate answers on the same examples and rubric. See scores, uncertainty, cost, and failures before choosing.	First task
Run a task consistently	Release a model with a task version and saved validation and routing settings. Inspect answers against that release.	Compare and release
Answer from your documents	Add task sources and inspect the passages retrieved for each answer.	Sources
Set what an agent can do	Check wrapped actions against permissions, limits, and human approvals. Inspect each run and its decisions.	First agent
Share instructions	Publish skill versions for employee AI tools and application agents. Track retrieval and declared use.	Team skills
Improve the next release	Review corrections and saved steps, add accepted examples to a dataset version, and compare again.	Improve from production
Review work across a team	Give people the access they need, route approvals, and inspect production and change records.	Review work

Applications connect through the task API, a text chat-completions endpoint, or the TypeScript and Python SDK integrations. Teams can use Orlo with an existing model gateway. Choose an integration.

For deployment options, see Operate Orlo. For the public SDK source and packages, see Open source.