Decide where AI runs, what it may see and who reviews its work before you switch it on for every team.
Choose where AI runs.
Pick a local Ollama host for data that must stay on your network, or a supported cloud provider for larger models. Application hosting and AI hosting are separate decisions, and cloud use for design from a file asks for consent first.
Agree what AI may see.
Explain data computes comparisons inside ReportBuilder360 and sends findings for wording. Prompts in the designer include the report context you are working on. Agree which datasets are in scope and record the decision.
Start with a period you know.
Run Explain data on a month or quarter your team already understands. Check that the drivers and key figures match what you expect before relying on the wording.
Review language and tone.
If readers use Arabic and English, review results in both. Decide which terms must stay fixed—product names, account names, regions—and check them in the output.
Keep the review step.
Treat every draft and summary as a proposal. Name who approves AI-assisted reports and dashboards before they are shared outside the team.