AI sequence prompts with constraints: specify evidence, structure and exclusions

Define the decision

An open request to 'write a high-converting sequence' invites unsupported claims and repetitive persuasion. A useful prompt tells the model what evidence it may use and what the recipient should be able to decide. It also creates criteria a reviewer can apply to the result.

Work through the procedure

1. List approved facts separately from unverified hypotheses. Include only evidence needed for the message.
2. Define the actual deliverable and one next action. State what must not be promised.
3. Assign a different contribution to each step and require missing facts to be flagged outside the sendable copy.
4. Review the output for invented details, repetition and unresolved placeholders before using it in the sequence editor.

Worked example

The following is a synthetic example for this procedure, not a customer result or performance benchmark.

Draft three messages for operations managers.
Approved fact: our offer is a one-page handoff checklist.
No customer metrics or recipient-specific facts are supplied.
Step 1: ask about ownership. Step 2: explain one checklist item. Step 3: close the thread.
Do not invent results, urgency or a prior relationship.
Return missing-information notes separately from email copy.

Read the result

The prompt constrains the model's evidence and narrative. It does not guarantee compliance, so the output still needs review. Keeping notes outside the copy reduces the chance that an instruction such as '[verify company fact]' is accidentally sent, but a final rendering check remains necessary.

Check before moving on

1. Compare every output claim with the approved-facts list.
2. Check that each step contributes something distinct.
3. Remove placeholders and editorial instructions from the saved message.

Limits and next action

This is a prompt pattern, not a promise of identical behavior across AI providers. Zintara's provider and fallback paths can differ. Use the same editorial acceptance criteria regardless of how the draft was produced.

Source: Zintara AI writing and review context