A client writes: “We need to update our website and get more enquiries. Please send a proposal.” Asking AI to produce polished copy immediately risks invented scope, budgets and deadlines. A better first step is turning the three lines into a reviewable decision structure.
1. Separate facts from unknowns
Open Yasnora AI and provide an anonymized brief you are authorized to use. Start with:
Divide this brief into confirmed facts, assumptions and unknown conditions. Show the source wording for every fact. Do not invent a budget, timeline, performance numbers or website platform.
In this example, only a desire to update the site and receive more enquiries is confirmed. The audience, current page performance, baseline enquiry volume and project constraints are unknown. A clear list of gaps is more useful than a premature quote.
2. Ask questions that change the scope
Ask Yasnora to select five questions that would materially change the proposal. Examples: what counts as an enquiry; which pages bring customers today; whether analytics data exists; what is wrong with the current site; and which time and budget limits apply. Record the decision affected by each answer.
| Question | Decision it affects |
|---|---|
| What counts as an enquiry? | Outcome definition and measurement |
| Is site analytics available? | Whether measurement setup is a first step |
| Which pages matter most? | First-stage priority |
Do not send a forty-question form by default. Five questions for the first discussion help the client and team define the work together.
3. Draft from agreed conditions only
After the client responds, add the confirmed facts to the conversation. Request a structure with the client's problem, stages, deliverable for each stage, client responsibilities, exclusions, open questions and next action. Require “to be agreed” wherever a condition remains unknown.
“Conduct an audit” is an activity. “Deliver a prioritized list of issues with examples and recommendations” is a reviewable outcome. The second formulation reduces ambiguity during approval.
4. Audit the promises before sending
Use a second prompt: “Identify unsourced numbers, unconfirmed promises, ambiguous dates and activities without deliverables. Do not silently rewrite them; show each risk and the clarification needed.” A person then checks the document against the correspondence, prices and the team's actual capacity. AI can prepare a draft; it does not approve commercial commitments.
This process only requires a chat with text you provide. Models and limits depend on plan and settings; files and external systems have separate eligibility. See the Yasnora capability status. Yasnora is part of Wicsora, alongside ProfitVena, which handles seller analytics. Technical part updated in ProfitVena.
A short Russian version on TenChat is also available.
