AI for sales has one clear aim: give selling time back to your reps. Today the average seller spends 40% of their time actually selling (Salesforce, State of Sales 2026), and the rest goes on researching accounts, entering data and writing emails. Those tasks around the sale are where AI saves hours without touching the customer relationship.
How do you use AI for sales?
AI for sales is used mostly to prepare and to follow up: researching an account before the meeting, summarising a call, updating the CRM, and writing the follow-up email or the first draft of a proposal. The conversation with the customer, the negotiation and the close stay with the rep.
Use is already widespread: 87% of sales organisations use some form of AI for tasks like prospecting, forecasting, lead scoring or email writing, and 54% of sellers say they have used AI agents (Salesforce, State of Sales 2026).
| Task | What the AI does | What the rep does |
|---|---|---|
| Preparing a meeting | Summarises public information on the account and the CRM history | Decides the angle and the questions |
| Calls | Summarises what was agreed and the next steps | Validates the summary before it is saved |
| CRM | Proposes field and note updates | Reviews and confirms the changes |
| Follow-up | Drafts the email | Adjusts the tone and sends it |
| Proposals | Prepares a first version on the company template | Validates the customer's challenges and the offer |
Sales · Three moments
What the AI prepares and what the rep decides.
Before
The AI prepares the account
Summarises public information and the CRM history. The rep decides the angle and the questions.
During
The rep has the conversation
The negotiation and the close are not delegated. The AI, at most, takes notes.
After
The AI writes the follow-up
A summary of what was agreed, the CRM update and a draft email. The rep validates and sends.
Source: Ironhack for Business.
What is the best AI for sales?
The best AI for sales is the one that connects to your CRM and your email, the one you already pay for in licences, and the one your team knows how to use. It can be Claude, Microsoft Copilot, ChatGPT or the AI your CRM already includes: the difference is usually less about the model than about the integration and the daily use.
Before buying a new tool, check three things:
- whether it can read your CRM and email data with the right permissions,
- whether you already pay for a licence that does the same thing,
- and whether your team can use it on their real tasks.
In our AI consulting we start with an audit of the tools already circulating in the company, unauthorised ones included, so you do not pay twice for the same thing. Sometimes building beats buying: when the licence for our meeting-notes tool expired, we built our own at Ironhack. It transcribes locally (the audio never leaves the laptop), recognises who is speaking, writes the notes, and connects to Claude so you can ask about any meeting. Licence cost for the team: zero.
If your company is weighing up providers, we compare the two most used assistants in Claude vs ChatGPT. And if the sales team works hand in hand with marketing, read our guide to AI in marketing too.
How we use it at Ironhack for Business: sales proposals with a Claude skill
In our own sales team we use a Claude skill, a reusable set of instructions, that prepares the first version of every proposal.
The flow has a fixed order: the AI reviews public information on the client, suggests possible challenges and puts them to the rep. Only once the rep has validated them does it generate the proposal on our template. If it is missing a fact, it asks instead of inventing one.
That rule is the heart of the system. A proposal with an invented fact about the client does more damage than a proposal that arrives a day late, so the rep still decides what goes in and what does not. The AI handles the mechanical part, and the judgement stays with the rep.
Why does AI in sales depend on clean data?
AI in sales depends on data because an agent can only be as accurate as the CRM it reads: with duplicates, empty fields or systems that do not talk to each other, the summaries and the recommendations come out wrong. More than half of sales leaders using AI (51%) say disconnected systems are slowing their AI initiatives (Salesforce, State of Sales 2026).
Which is why 74% of sales professionals are focusing on data cleaning, and high-performing teams go further: 79% prioritise data hygiene, against only 54% of low-performing teams, according to the same report.
Before automating, define which CRM fields are mandatory, who maintains them and how duplicates are removed. It is the least visible work in the project and the one that most decides the result.
How do you train a sales team in AI?
A sales team learns AI by practising on its own accounts and tasks, not on generic examples. The lack of practice already weighs on younger profiles: among Gen Z sellers, 46% rarely get feedback on their sales conversations and 47% do not have enough practice opportunities before talking to customers (Salesforce, State of Sales 2026).
Our AI School includes a Sales track at three levels: foundational (using AI well), intermediate (building with AI) and advanced (running it in production). The tracks are built from each company's diagnosis, with real use cases that have an owner, a validated prompt and a KPI, and more than 50% of the time is practice.
To size the impact of AI for commercial teams, our indicative estimate is 190 to 420 hours a year per rep, calculated over 47 working weeks. The real saving depends on task volume, adoption and integration, and you can estimate yours with the savings calculator.
Where to start with AI for sales: 4 steps
- Pick one task. The one that takes the most time and repeats the most, such as preparing meetings or writing follow-ups. Measure how long it takes today.
- Run a trial on real data. Start small: give an agent a month of sales data to see what the weekly report would look like. Leave full CRM integration out of this first phase.
- Train the team that will use it. Our hands-on workshops can be run per department, on Claude, Copilot or ChatGPT depending on the tools your company already has, and end with commitments to apply what was learned, signed in the room.
- Measure before you scale. Compare the hours the task took before and 60 to 90 days later. If the result improves and the errors are under control, move to the next task.
Next step
If you want to know which task in your commercial team is the best candidate to start with, we will come back with a starting point in under 1 working day.