Improve Omni AI answers with model context
Teach Omni’s analytics assistant what team terminology means using field descriptions, topic context and examples that can be tested on a separate branch.
To improve Omni AI answers with model context, identify a question the assistant misunderstands today. A sales team might use “new accounts” for first-time paying customers, while the model also contains every newly created account. The omni-ai-optimizer skill helps express that distinction where the assistant selects topics and fields.
The change might be a field description, a synonym, a short topic instruction or a representative example. The workflow encourages testing on a branch so the team can inspect the effect on neighbouring questions before updating the shared model.
What it does
- Map common business terms to the appropriate dimensions or measures.
- Clarify ambiguous descriptions at model, topic or field level.
- Provide examples based on questions colleagues actually ask.
- Review settings that shape the assistant’s available context.
- Compare answers before and after the proposed modelling change.
Install and use
- Install the skill and official Omni CLI. Use an authenticated profile with model editing and branch access. Testing the hosted assistant requires the appropriate Omni AI capability.
- Ask, “The phrase active customer should use our 30-day purchase definition. Find the right field and propose the smallest context change that would help.”
- Test the original question and a related question that should still behave differently, then review the branch.
FAQ
Source
Made by Omni Analytics, Inc. Original skill: omni-ai-optimizer, published on GitHub under the Apache-2.0 licence.
View on GitHub