AI Assistant in Sales: What It Does and What It Does Not

A salesperson spends a surprising share of the day not selling: copying details between systems, hunting for the last conversation with a customer, assembling a quotation from a template. That part transfers. The conversation itself does not.

What does an AI assistant do for a salesperson?

• Sorts the incoming. Reads enquiries from every channel, works out what is being asked, creates the record and puts it in front of the right person with the substance summarised.
• Prepares the customer record before a call. Previous correspondence, what was agreed, what was quoted, what is outstanding — collected into a short brief rather than five open tabs.
• Drafts the routine documents. Quotations, specifications, standard contracts filled from the record, ready for a person to check and send.
• Keeps the follow-ups. Who promised what and by when, which customer has gone quiet, which quotation has had no answer for a week.
• Writes up the call. Turns the notes into a record entry so the next person to touch the account is not starting from nothing.

What stays with a person?

Everything involving judgement about money and trust. Discounts and terms. Objections, where the real reason is rarely the stated one. Any promise about a date, a scope or an exception. Difficult conversations after something has gone wrong.

The line is simple to state: the assistant prepares and records, the person decides and commits. Cross that line and you get a system that confidently promises a customer something the company cannot deliver — which costs far more than the time it saved.

How do you know whether it pays for itself?

Measure one thing before and after: hours per week your sales team spends not talking to customers. Add up the record-keeping, the searching, the document assembly, the copying between systems.

Typical outcome for a small team: six to ten hours a week returned per salesperson. At that scale the arithmetic is usually straightforward. If the recovered time is under two hours a week, the process was not the bottleneck and the money is better spent elsewhere.

The second measure, slower but more telling: how many enquiries get a reply within an hour, and whether that share rises after launch.

What are the common mistakes when introducing one?

• Letting it answer customers directly from day one. Start with preparation and record-keeping, where a mistake costs nothing. Customer-facing comes later, once you trust the output.
• Introducing it without the team. If salespeople read it as a threat to their jobs, it will not be used, whatever it does. Show it as removing the part of the job nobody enjoys.
• No agreed line of responsibility. Write down explicitly what the assistant may decide and what it must escalate — before launch, not after the first incident.
• Feeding it a disorganised CRM. An assistant working from records that are half-empty and inconsistent produces confident nonsense. Tidy the data first.

Where should you start?

With the least interesting task your salespeople do every day — the copying, the searching, the assembling. Measure how long it takes across a week, hand that one job over, and check the hours again a month later. That single number decides whether to extend it further.