AI in sales is the use of machine learning and generative AI to research accounts, write first drafts, take notes, score calls and forecast deals, so sellers spend more of their time talking to buyers. It works best on the work around the conversation, and least as a replacement for the conversation itself.
This guide maps where AI fits at each stage of a B2B deal, with examples, what should stay with people and how to start with a team. It is part of our guide to sales techniques, which covers the moves AI is meant to support.
What is AI in sales?
AI in sales is any software that learns from data to do or support a sales task. Three kinds show up in most sales tools today, and they do different jobs.
| Type | What it does | Sales examples |
|---|---|---|
| Predictive AI | Finds patterns in past data to score or forecast | Lead scoring, deal risk flags, forecasts |
| Generative AI | Writes or summarizes text, audio or images | Email drafts, call summaries, account briefs |
| Conversational AI | Talks or listens in real time | Chatbots on the website, live call notes, AI role play buyers |
Predictive AI has been in CRMs for years. What changed recently is generative AI: tools that can read a call transcript and write the follow-up, or play a skeptical CFO in a practice call. That shift moved AI from the sales operations team into each seller's day.
Adoption is wide. In Salesforce's 2024 State of Sales survey of 5,500 sales professionals, 81% of sales teams said they were experimenting with or had fully implemented AI, and 83% of teams with AI saw revenue grow that year, against 66% of teams without it. Those numbers show that teams with AI grew more often. They don't prove that AI caused the growth, since teams that adopt new tools early may differ in other ways.
Where does AI fit in the sales cycle?
AI helps at every stage of a deal, but in a different way at each one. Before the call it saves research time, during the call it takes notes, and after the call it writes, scores and flags. The map below shows where it fits and what stays with the seller.
Prospecting
Account research, lists, first email drafts.Discovery
Prep questions, live notes, the summary.Demo
Examples from past calls that match this buyer.Objections
Practice with an AI buyer before the real call.
Closing
Follow-up email, mutual plan draft, deal risk flags.
Coaching
Call scores, patterns across calls, practice plans.
Here is the same map with what to hand to AI and what to keep.
| Stage | Hand to AI | Keep with the seller |
|---|---|---|
| Prospecting | Research on the account, the trigger, the people | Choosing the angle and sending the message |
| Discovery | A list of likely questions, notes during the call | Asking, listening and following the answer |
| Demo | Finding the right customer example | Reading the room and changing course live |
| Objections | Playing the buyer in a practice call | Handling the real objection with the real buyer |
| Closing | Drafting the recap and next steps | The ask, and the relationship with the champion |
| Coaching | Scoring every call against the playbook | The one-on-one conversation about what to change |
The pattern: AI takes the tasks that come before and after a decision. People keep the decisions and the conversations.
What are examples of AI in sales?
Here is what AI looks like in the week of one account executive at Corvane, a fictional logistics software company. Each example is a real use you can set up with tools that exist today.
Research before a first call
Before a call with a 200-person freight company, she asks an AI assistant for a one-page brief: what the company does, recent news, open roles in operations and the likely priorities of the VP she is meeting. It takes two minutes instead of 30. She reads it, checks two facts on the company's site and picks one angle for her opening.
A first draft of the outreach email
She gives the AI the trigger (the company just opened a new warehouse) and asks for a four-line email. The draft is generic, so she rewrites the second line with something only she knows from a similar customer. The AI saved the blank page. She wrote the part that gets a reply.
Notes and a summary after discovery
A note-taker joins the call, writes the transcript and drafts a summary with the pains, the people mentioned and the next step. She fixes one wrong name and sends the recap to the buyer within the hour.
Practice before a hard call
Her next call is with the CFO, who has pushed back on price before. She runs a ten-minute practice call with an AI buyer set up to play a cautious CFO, tries two ways of framing the cost and keeps the one that sounded more natural.
Feedback on every call
Her calls are scored against the team's discovery scorecard. Her manager sees that she skips the question about who else is involved in four calls out of five and brings one of those calls to their next one-on-one.
A flag on a deal going quiet
The CRM flags a deal where the buyer hasn't replied in 12 days and no next meeting is booked. She calls her champion and learns about a re-org that would have killed the deal if she had waited another week.
What should stay with people?
AI is good at the steps around a conversation and weak at the conversation itself. Buyers can tell when a message was written for everyone, and they decide whether to trust a seller in the moments AI can't fake: a good follow-up question, an honest "we're not the right fit for that", a call where the seller clearly listened.
Hand to AI
Tasks with a clear input and output:
- Research and account briefs
- First drafts of emails and recaps
- Notes, transcripts and summaries
- Scoring calls against a scorecard
- Playing the buyer in practice
Keep with people
Moments that build trust and need judgment:
- Choosing which accounts and angle to pursue
- Asking, listening and following up live
- Handling a real objection
- Asking for the decision
- Coaching conversations between a manager and their team
A useful test: if a buyer found out AI did this step, would they mind? Nobody minds that a seller used AI to research their company. Many would mind finding out their "personal" email was written for 500 people at once.
How do you start using AI in your sales team?
Start with one task that eats time and has a clear output, run it with a few people for a month and measure the time saved before you add the next. Teams that roll out five AI tools at once usually end up using none of them well.
Pick one task
Call notes and recaps are the usual first choice.
Write the rules
What data the tool can see, and what a person must check before it goes out.
Pilot with a few people
Three to five people, for four weeks.
Measure the time saved
Hours a week per person, and whether quality held.
Add the next task
Only when the first one is a habit.
Good first tasks, in the order most teams add them:
- Call notes and recaps. Low risk, saves time on every call, and creates the transcripts the next steps need.
- Account research. Saves time before every first call, and a person reads the brief before using it.
- Email first drafts. Useful once people treat the draft as a draft.
- Call scoring. Turns the transcripts into feedback for every call instead of the few a manager can listen to.
- AI role play. Lets people practice a hard conversation before it happens.
Go deeper: how to use AI in sales walks through five of these uses step by step, with an example prompt for each.
The time you save is only worth something if it goes back into selling. Our guide to sales productivity covers how to measure where a seller's week goes and protect the hours for real conversations.
How is AI changing the work of a seller?
AI is moving routine sales work from typing to checking. Gartner predicted in 2023 that by 2028, 60% of B2B seller work will be executed through conversational user interfaces via generative AI, up from less than 5% in 2023. A prediction is not a measurement, but the direction matches what teams see: fewer hours writing notes and emails, more hours reviewing what the AI drafted.
That changes which skills matter most. When AI handles the first draft, the seller's value is in what AI can't do: understanding the buyer's problem, asking the next question and earning the trust to ask for a decision. Our guide to sales skills covers those skills in the order a deal asks for them.
What are the risks of AI in sales?
The risks of AI in sales are mostly about trust: trust in the data, trust from the buyer and trust inside the team. Each one has a simple guard.
How is AI in sales different from AI sales coaching?
AI in sales covers every use of AI across the deal, from research to forecasting. AI sales coaching is one part of it: using AI to score calls, find the skills each person should work on and give them a way to practice. It is the part aimed at making people better at selling.
If coaching is the problem you are trying to solve, our guide to AI sales coaching covers what it is, how it compares to traditional coaching and how to run it with your team. For a session on the basics, watch intro to AI selling, and for a look at the tools, powerful AI tools for sales.
Where to start
Pick the task that takes your team the most time around each call, usually notes and recaps, and pilot one tool for a month with a few people. Keep the conversation with people, check every fact the AI produces and put the time you save back into selling. For the techniques that make those conversations count, go back to our guide to sales techniques.