How to use AI in sales training

AI can do three useful jobs in a sales training programme. It can capture what your best salespeople know, let everyone on the team look it up whenever they need it and give people realistic sales conversations to practise between training sessions. It works best alongside trainer-led training, because people change how they sell through practice with feedback.

This article is for sales directors and for people who buy learning and development for their company. It covers the three jobs, what the research says about practice and feedback, where AI fits beside a trainer, what it cannot do and what to ask a provider.

Inner Leader trains B2B sales teams. We are working on this now with a specialist AI partner, alongside our consultative sales training. This is current work, so the evidence in this article comes from published research and none of it is a result of our own.

The three jobs AI can do in a sales training programme

Using AI in sales training means using AI software to help salespeople learn and keep the skills a training programme teaches. It does three jobs: it records what your best people know, it makes that knowledge available to everyone when they need it and it gives people a realistic conversation to practise on.

1. Capturing what your best salespeople know

Top performers often carry answers that are not written down anywhere. They know how to respond to the hardest questions, what to say when a buyer pushes back on price and how to open a pitch. The first job is to interview them and record those answers. Our guide to capturing what your top salespeople know explains what to ask in the interviews and why experts leave things out when they describe their own work.

2. Letting every salesperson look up that knowledge when they need it

The recorded answers can go into an AI assistant, which is software a person can ask a question in plain English. It is set up to answer from your own team's material. A salesperson can type a question before a call, such as how to respond when a buyer says a competitor is cheaper, and read how your best people handle it.

The strongest evidence we have found comes from customer support. In The Quarterly Journal of Economics in 2025, Brynjolfsson, Li and Raymond studied 5,179 support agents who were given an AI assistant. Issues resolved per hour rose by 14% on average and by 34% for novice and low-skilled workers, with minimal impact on experienced workers.

The authors found suggestive evidence that the tool was passing on the best practices of the more able workers. The study did not cover sales teams.

3. Letting people practise between training sessions

The third job is practice. A salesperson holds a realistic sales conversation with an AI that plays the buyer, built from the questions and objections your own customers raise. They can do it between workshops, as often as they like.

Simulation-based learning means learning by practising in a realistic situation that has been set up for the purpose. A meta-analysis by Chernikova and colleagues in the Review of Educational Research in 2020 pooled 145 studies of it in higher education. It found a large positive effect on learning. We have not yet found an equivalent body of peer-reviewed research on AI role-play for sales teams.

Objections are a good place to start practising. Our guide to sales methodologies and qualification frameworks shows how an objection often points to a stage of the conversation that was not finished.

Why practice with feedback changes how people sell

People change how they sell by practising and being told specifically what to do differently. Anders Ericsson and his colleagues (1993) used the term deliberate practice for effortful activity designed to improve performance. A habit changes when a salesperson uses a new approach in a conversation, hears how it went and tries again.

What happens when people practise with no feedback

A study presented at the CHI conference in 2026 tested this with AI. Ninety-four novice counsellors practised with patients played by an AI. One group practised alone and the other also received feedback on their responses. The group with feedback improved their use of reflections and questions. In the group that practised alone, empathy declined.

The authors concluded that “combining simulated practice with structured feedback is critical for meaningful improvement”. The study was in counselling, and we think the same caution applies to a sales team left alone with an AI role-play tool.

The feedback itself has to be good. Kluger and DeNisi's 1996 meta-analysis in Psychological Bulletin pooled the results of many earlier studies of feedback. It found that feedback improved performance on average, and that in more than a third of the cases they studied the feedback made performance worse. For that reason we recommend that a trainer or manager stays involved in what an AI tool tells your salespeople.

Why practice between sessions helps people remember

A review of 317 experiments by Cepeda and colleagues found that spreading learning out helps people remember more than cramming.

Roediger and Karpicke (2006) found that students who were tested on material remembered more of it two days and a week later than students who studied it again. Short practice conversations between workshops spread the learning out, and they make people recall what they learned.

Where AI fits beside trainer-led sales training

AI is most useful between and after the live sessions, and a trainer or manager should stay involved in the feedback. The clearest evidence is a set of field experiments, which are experiments run with people doing their real jobs. They were published in the Journal of Marketing by Luo, Qin, Fang and Qu. In the first, 429 sales agents were randomly given an AI coach or a human coach.

The AI coach helped middle-ranked agents most. Bottom-ranked agents were overloaded by the amount of feedback and top-ranked agents were the most resistant to feedback from an AI. In a later experiment, human managers passed on the AI's feedback, and both bottom-ranked and top-ranked agents did better than with either coach alone.

What AI cannot do in sales training

A provider should be willing to name the limits of its AI tools. These are the five we would raise with any sales director.

• It can give a wrong answer confidently. A study in the Journal of Empirical Legal Studies tested legal research tools built to answer from trusted legal sources. The authors reported that the tools “each hallucinate between 17% and 33% of the time”. Hallucinating is the term for an AI tool making something up. We suggest you name one person to check what your assistant says about prices, contract terms and product claims.
• It only knows what it was given. It will not know about a new product or a new competitor until someone adds it.
• People can lean on it and stop learning. In a study of nearly a thousand high school maths students published in PNAS in 2025, students with a standard AI chat tool scored 48% better in practice sessions. Once the tool was taken away they scored 17% worse than students who never had it. A version designed with safeguards largely avoided the harm.
• It does not replace a manager or trainer who knows the salesperson. In the Journal of Marketing experiments, the lowest-ranked and highest-ranked agents did better when a human manager delivered the AI's feedback.
• It raises confidentiality questions. Recorded interviews contain personal data and often customer names and prices, so you need to know where that material is stored and who can see it.

What to ask a provider of AI sales training

These six questions will tell you most of what you need to know about a provider.

• Whose knowledge is the assistant built on: our own team's answers or general material from the internet?
• Who checks the answers before our salespeople see them, and how are wrong answers corrected?
• What feedback does a salesperson get after a practice conversation, and where does a trainer or manager come in?
• Where is our data stored, who can see it and is it used to train an AI tool that anyone else uses?
• How are the people you interview told what the recordings are for, and can they withdraw?
• What results can you show, and from which clients?

How Inner Leader is approaching AI in sales training

Inner Leader is working on this now with a specialist AI partner. The work starts with interviewing the top talent in a sales team, which can be a global sales team. Their answers become the team's own source of wisdom and knowledge: how to answer difficult questions, how to handle objections and how to pitch and win business.

That same source can then become a sales training simulation that supports the in-person or virtual sales training we deliver. Our training is built on PEPITA®, our six-stage consultative selling model, which people practise in role-plays on their own products and buyers.

This is current work and we do not have results of our own to share yet. If this is something you are exploring for your own sales team, you can tell us about your sales team and we will reply within one business day.