Find an answer
Choose the question that matches your next step. Each guide starts with the answer, then explains how to act and what to verify. For the overall request paths, start with How Orlo works.
Get started
- How do I get my first checked task answer?
- How do I compare support-routing models with Orlo?
- How do I replay an evaluation without a model key?
- How do I verify an agent connection?
- How do I use a shared skill in my AI app?
Tasks and production answers
Shared skills and access
- How do I import an existing skill?
- How do I publish shared instructions?
- How do I use a skill in a task release?
- What happens when I retire a skill version?
- How do I invite someone with limited access?
- How do I create a key for my application or agent?
- How do I add or replace a model provider key?
- How do I choose the model that judges a comparison?
- Can I use my organization’s skills in a personal AI account?
- Why can’t I see a console page or action?
- Why is my AI app connected but not using the skill?
- How do I reopen Getting started?
Applications and agents
- How do I reuse answers for repeated task requests?
- Can my application keep streaming model responses?
- What happens to an agent when Orlo is unreachable?
- What happens when an answer fails a rule?
- What happens when a model provider fails?
- What should my application do when Orlo is busy?
- How does my application resume an approved agent action?
- What happens when a webhook cannot be delivered?
- How do I send Orlo traces to my observability service?
Measured permissions and learning
- How do I reduce repetitive approvals without granting broad access?
- What happens before a permission has enough evidence?
- Why did automatic permission pause?
- How do I review an action independently?
- What do Orlo’s confidence bounds mean?
- How do I identify an agent configuration for measured permissions?
- How do I report whether an agent action succeeded?