Preparing quotes with AI: the machine drafts, you decide

The most time-consuming part of preparing a quote is gathering the information. What did the customer ask for on the phone last week, which product did they ask about, which item was removed from the previous quote? If that information is on the record, AI can read it and turn it into a draft. The salesperson starts from a text to correct rather than a blank page.

The second contribution of AI is language. Quotes written by different salespeople for the same product drift apart in tone, detail, and layout. An AI working from a company-approved template and sample quotes builds every draft in the same structure, so the customer feels they are dealing with one company whichever salesperson sends the quote. The template itself is still human work and is reviewed regularly.

The quality of the draft depends on the quality of the record. If meeting notes were never written or the product list is out of date, the AI fills the gaps with guesses. So the first condition for preparing quotes with AI is keeping the basic fields of a quote in order on the record.

Under Article 3 of the Turkish Code of Obligations, a person who proposes a contract and sets a time limit for acceptance is bound by the offer until that time runs out. A quote sent at the wrong price can bind the company to that price for its whole validity period. The Personal Data Protection Authority's recommendations on artificial intelligence also ask that the role of human intervention be established in decision-making.

When a salesperson pastes a customer's name, phone number, and earlier bargaining notes into a general chat tool, that data leaves the company's record-keeping. If the prompt goes to an AI provider, that provider becomes part of the data processing chain. What a quote needs is product, quantity, scope, and terms; the customer's home address or personal notes are not needed for a draft.

Example scenario: picture a company in Gaziantep selling packaging machines. If a salesperson preparing a quote for a food producer met at a trade fair gave the AI the fair notes, the machine's capacity options, and a similar quote from last year, they would get a good draft. The personal phone numbers mixed into those notes and a competitor's price, on the other hand, would be removed before anything was shared.

Writing prompts has its own discipline. A good prompt states the purpose of the quote, the customer's question, the template to use, and the fields not to touch: do not change the price, do not add items, take the terms from the template. A vague prompt produces a vague draft, and the time spent checking it takes back the time saved.

• Does every item rest on something the customer asked for, or did the AI copy it from a similar quote?
• Do the prices come from the current price list?
• Is the discount within the salesperson's authority?
• Are the payment, delivery, and validity terms the same as the company template?
• Do the customer's name, title, and contact details belong to the right person?
• Does the text contain a promise that is not in the offer, such as free installation?

Quotes are most often lost in the silence after they are sent. The AI's job here is to bring the silent quote to its owner's attention. Sending the follow-up to the customer by itself is a separate decision, for the relationship and for consent; commercial messaging rules apply at this step too. This article is not legal advice.

X AI is the AI assistant that works inside Corex360. It does not send a reply or a quote to a customer on its own; it prepares a draft, and sending happens with approval. It stops and asks at money, contract, and people decisions; sending, discount, payment, leave, and permission steps wait for approval. It reminds you about quotes that have had no reply. A quote is built from a template on the opportunity card: product, quantity, price list, and terms are the template's fields.

In Corex360 Cloud, the company chooses the AI provider in the X AI settings and enters its own key; in that case the prompts written to X AI go to the chosen provider. In an on-premise installation, X AI works with a local model, and the prompts and the records it reads do not leave the customer's server. You can see the whole quote flow on the quote tracking software page.