Use and govern Qintorava AI
Review cited evidence, validate proposed actions, and trace agent configuration and runtime decisions.
Demo baseline · November 13, 2026Use assistance with record context
Open the portal or Agent workspace assistant, or the contextual assistance for the relevant record. Ask for the goal and include the record type or identifier when appropriate. Review the returned evidence, scope and proposed action. Use the visible review action and confirm the target and changes before applying a mutation.
Configure an agent
- Open the authorized AI control plane or Agent Designer surface. Identify the agent, module, permitted tools, and instruction version.
- Test scoped examples with known expected outcomes and unauthorized or incomplete inputs. Check retrieval evidence, proposed operations, confirmation behavior and execution receipts.
- Submit the draft for the configured review. Publish through the native workflow and verify the effective version. Review runtime logs and evaluation outcomes.
- When a version needs reversal, use the approved revision workflow and verify that the effective agent and guardrails match the intended baseline.
Troubleshoot AI output
If context is absent, check record access, active module and retrieval coverage. If the answer is incorrect, compare the cited sources and underlying record fields. If an action was refused, inspect the configured tool, required confirmation, current record state and authority. Capture the run ID, agent version and error without sharing hidden prompts or sensitive payloads.