Give AI price information without inventing a quote

AI can explain approved price information without issuing a custom quote. Give it a current reference with the service, price basis, included scope, exclusions, and conditions. When the customer's situation needs judgment or the reference does not answer it, have the system collect the question for a person instead of calculating a plausible total.

The difference matters. "This service starts at a published amount under these conditions" explains a reference. "Your project will cost this amount" may create a commitment your business has not reviewed.

Build a price reference people can check

Start with the information the business actually approves for public use. A useful record identifies:

• The service or product.
• Whether the figure is fixed, starting, recurring, or an estimate range.
• What that figure includes and excludes.
• Conditions that change whether it applies.
• When it was approved and who maintains it.
• The next step when the request falls outside it.

Do not use an old proposal as the universal price list. A past customer's quote may reflect different scope, materials, discounts, or timing. It is also not automatically public information.

Keep the approved record in one known place. If the website, assistant reference, and sales notes disagree, resolve the conflict before letting the assistant answer.

Preserve the conditions in the response

Keep "starting at" attached to the starting price, and "per month" attached to the monthly fee. Dropping those words changes what the customer thinks you offered.

Ask the system to carry the relevant qualifier into the answer, not hide it at the end of a long explanation. It should distinguish "included in this published package" from "available as separately scoped work."

A fictional example: a service business publishes a starting amount for a routine visit, while repairs depend on assessment. The assistant can explain the approved visit information. It should not estimate the repair from a short description or a blurry photo.

For LeadSpark, the current services page separates Business Hub website plans from CRM connections, follow-up, scheduling, and other custom work through a System Audit. An assistant should not add those systems to a website plan because the customer asks about them.

What if the price record is out of date?

Stop using it for answers until someone confirms the current information. An approval date helps identify what needs review; it does not guarantee the figure is still valid.

Review the reference after a price or scope change, and test ordinary questions again. Check whether the system still repeats a retired offer or a condition you removed.

Can AI use a formula to prepare an estimate?

A defined calculation may support an internal estimate when the inputs and rules are reliable. That needs its own scope, testing, and approval boundary. Do not let a language model improvise the formula or fill in missing measurements.

Keep the estimate's assumptions visible. A person should confirm anything that affects the customer's final price or commitment.

Test the requests that invite guessing

Ask about a discount, an unusual project, an omitted condition, an old price, and "What is my total?" Review whether the system cites the approved reference or reaches the human review step.

Our guide to choosing AI or ordinary automation explains why fixed calculations and business judgment belong in different parts of a workflow. The practical AI guide covers choosing a narrow first task.

If price information crosses several tools or needs an approval process, start a System Audit conversation. Bring the current reference and the exceptions. Clear pricing support begins with an approved rule, not a confident answer.