Use these tips to write effective prompts for Zilliant MCP workflows and get clear, data-backed answers grounded in your pricing logic.
For best results:
Include the identifiers Zilliant needs
Your AI client can return a useful, verifiable answer only if it can identify what you are asking. Include the identifiers your organization uses in Zilliant products, such as:
Customer name or customer ID
SKU, product ID, or item group
Requested date or date range
Region, segment, or channel
If you leave identifiers out, the workflow may fail or the answer may be incomplete.
Ask for the structure you want
If you plan to reuse the output, ask for a structured response.
Examples:
“Return the explanation as: Recommendation, Top drivers, Evidence (data points), Guardrails/policy checks, and Next action.”
“List the top 3 drivers behind the recommended price, such as cost, customer segment, volume, and discounts. Provide supporting evidence for each.”
“Include the Price Waterfall breakdown (list each step and the value/impact) and call out any overrides.”
Keep each prompt focused
Ask one question at a time. Prompts that combine multiple questions or many products may be slower, less consistent, and harder to audit.
If you need to review multiple items, start with one, confirm the result looks right, then reuse the same prompt pattern for the rest.