SALES · PRACTICAL GUIDE

How to Use AI in Sales: A Practical Guide for Reps and Sales Managers

Ask a sales team where their week goes and the answer is rarely “selling”. It goes on researching accounts, preparing for calls, writing proposals, chasing follow-ups and updating the CRM. AI can take a real share of that load, but used carelessly it also produces generic emails, invented facts about prospects and a pipeline nobody trusts. This guide shows how to use AI in sales step by step, from first research to pipeline review, with prompts you can copy, a worked example and the guardrails that protect your reputation with buyers.
By the Skillopedia team  |  Updated 19 October 2026  |  15 min read

What AI is good at in sales, and what it is not

Selling is mostly reading, writing and preparing: annual reports, LinkedIn posts, call notes, emails, proposals and CRM updates. AI assistants such as ChatGPT, Microsoft Copilot, Claude and Gemini, along with the AI features now built into most CRMs, are quick at all of that. They can summarise a 200-page annual report into what matters for your pitch, turn rough call notes into a clean recap and produce a first draft of a proposal section in minutes.

What they can’t do is know your buyer. They don’t know that the CFO was burnt by a failed rollout two years ago, that procurement only buys through a framework agreement, or that your champion is about to move roles. Left to themselves, they also fill gaps with plausible guesses, which is how a rep ends up congratulating a prospect on an acquisition that never happened. The working rule:

  • AI researches, prepares, drafts and tidies.
  • Reps build the relationship, verify the facts and make every commitment.
Where AI helps across the sales cycle: research, account plans, call briefs, proposals, follow-ups and pipeline review
Six places in the sales cycle where AI saves the most time.

Before you start: four things to agree

  1. Which tool. Use the AI tool your company has approved on a work account, or the AI features in your CRM if they have been switched on. Don’t paste customer contracts, pricing or contact lists into a personal chatbot.
  2. What data may go in. Company information is usually fine; personal data about contacts is regulated. India’s Digital Personal Data Protection Act and the UAE’s Personal Data Protection Law both apply to prospect and customer details, so use what the task needs and nothing more.
  3. What stays human. Pricing, discounts, contract terms, delivery promises and anything said in the customer’s name.
  4. A baseline. Note how long reps spend on call prep, proposals and CRM updates, and your proposal turnaround time, before you change anything.

New to AI at work? Our guide to AI literacy training for employees covers the basics every rep should know first.

Nine sales workflows, step by step

Each workflow has a prompt you can adapt. Replace the brackets, paste only real sources and read every output before it reaches a customer or the CRM.

1. Account research from real sources

The biggest mistake is asking an AI “tell me about Company X” and trusting the answer. Instead, give it the sources and ask it to work only from them.

Prompt: I sell [what you sell] to [type of customer]. Using only the documents I have pasted below (latest annual report, two recent press releases and the “About us” page), summarise: the company’s stated priorities for the next year, any expansion, cost or technology initiatives, named leaders relevant to [your area], and risks they mention. Give the source for every point. Then list three hypotheses about where we could help, clearly labelled as hypotheses.

If your approved tool has web search, you can ask it to find sources too, but open the links and check them yourself before using anything in a conversation.

2. Account plans for key customers

For strategic accounts, AI can turn a scattered set of notes into a structured plan the whole account team can work from.

Prompt: Here are our notes on [account]: products they use today, contract renewal date, contacts we know and what each cares about, open opportunities, and recent support issues. Build an account plan with these sections: relationship map (with gaps), current value delivered, whitespace (products or divisions we don’t serve yet), risks to the renewal, and a 90-day action list with an owner for each action. Flag any section where the notes are too thin to be useful.

3. A one-page call brief

Good call prep is the habit that separates strong reps from busy ones. AI makes it quick enough to do for every call, not just the big ones.

Prompt: Prepare a one-page brief for a first discovery call tomorrow with [roles of attendees] at [company]. Use the research summary and CRM notes below. Sections: why they might be talking to us now, who is in the room and what each probably cares about, three hypotheses to test, six discovery questions, likely objections, and the outcome I should aim for. Label anything inferred as an assumption and write “unknown” where information is missing.
Example of a one-page sales call brief prepared with AI
A call brief like this takes minutes with AI, and every line can be checked.

4. A discovery question bank by persona

Ask a team to share their best discovery questions and you will get a dozen documents. AI can turn them into one bank, organised by buyer role and deal stage.

Prompt: Here are discovery questions collected from our top reps, plus our qualification checklist. Organise them into a question bank for three buyer roles: [operations head, IT head, finance controller]. For each role give eight open questions grouped as current situation, pain and impact, decision process, and timing. Remove duplicates, rewrite any leading or yes/no questions, and add follow-up probes for the four most important questions.

The output is a draft for your sales manager to approve, then a shared asset the whole team uses.

5. Same-day recaps and follow-ups

A clear recap within hours of a meeting is one of the simplest ways to look professional and keep deals moving. AI can turn your notes or an approved meeting transcript into one.

Prompt: From these call notes, draft a follow-up email to [name] that thanks them briefly, summarises what we heard in their words (three bullets), confirms the agreed next steps with owners and dates, and attaches nothing we didn’t promise. Keep it under 150 words, warm but businesslike. Separately, list anything in the notes that sounds like a commitment from our side, so I can check it with my manager.

That second instruction catches the “we can probably do that by month-end” remarks that turn into problems later. If you record calls, follow your company’s consent policy and local law; tell participants when a meeting is being recorded or transcribed.

6. Proposals and RFP responses

Proposals are where AI saves the most hours, and where careless use does the most damage. The safe pattern is to work from your own approved content, not the open internet.

Prompt: Attached are the client’s RFP, our standard proposal template and three approved case summaries. First, build a compliance matrix listing every requirement in the RFP with the section where we will answer it. Then draft the executive summary (300 words) tying the client’s stated objectives to our approach. Use only capabilities described in the attached documents. Mark any requirement we can’t evidence from our material as “Needs input from [team]”.

Every claim about delivery, timelines, security or pricing must be checked by the person who owns it. AI drafts the words; it can’t sign up your delivery team to a promise.

7. Objection handling practice

AI makes a patient role-play partner. Reps can rehearse tough conversations before they happen, without taking a manager’s time for every practice round.

Prompt: Play the role of a sceptical finance controller at a mid-size manufacturer who has been asked to approve our software. You think the price is high, you are happy with spreadsheets, and you worry about implementation time. Raise one objection at a time and respond realistically to my answers. After six exchanges, step out of role and give me feedback on what worked, what sounded defensive and one better way to answer each objection, based on our playbook pasted below.

Want the whole team selling this way?

One rep with good prompts is a nice story. A team sharing a prompt library built on your own accounts, playbook and CRM fields is a change in how you sell. Our sales programme has reps practise these workflows on live (masked) deals, with managers agreeing the guardrails first.
See the sales programme
Talk to our team

8. CRM notes that people actually trust

Most CRMs are only as good as the notes reps find time to write. AI can turn rough notes into structured updates, and point out what is missing.

Prompt: Turn these rough notes into a CRM update with the fields: summary (two sentences), stakeholders met and their roles, pain points in the customer’s words, budget status, decision process, timeline, next step with date, and risks. Write “not discussed” for any field the notes don’t cover. Then list the three most important questions I still need answered to qualify this deal properly.

Many CRM platforms now include AI features that summarise emails and calls into the record. The same rule applies: the rep reads and corrects the summary before saving it.

9. Pipeline reviews for sales managers

Managers can use AI to prepare for a pipeline review rather than run it. Give it an export of open deals and ask where to look first.

Prompt: Here is our open pipeline export with stage, value, close date, last activity date, next step and notes. Flag deals where: the close date has moved more than once, there has been no activity for 21 days, the next step is missing or vague, or the notes show no contact with a decision-maker. For each flagged deal, suggest two coaching questions I could ask the rep. Do not change any forecast figures.

Use the flags to start conversations, not to judge reps automatically. The rep usually knows something the export doesn’t.

Worked example: preparing a discovery call in 20 minutes

Here is the call-brief workflow in practice for an illustrative case: a rep selling planning software, preparing a first call with a mid-size logistics group in the UAE and Oman.

  1. Gather sources (5 minutes). The rep saves the group’s latest press release about a new warehouse, its careers page showing a transformation lead vacancy, and the CRM note from an inbound enquiry.
  2. Run the research prompt (2 minutes). The summary lists the warehouse expansion and the hiring push, each with its source, and three hypotheses including manual dispatch planning across sites.
  3. Run the call-brief prompt (2 minutes). The brief suggests the COO is the likely sponsor, the IT head an evaluator and the finance controller the budget holder, with six discovery questions.
  4. Check and correct (8 minutes). The rep spots that the AI described the company as “recently acquired”, which no source supports, and deletes it. She swaps one generic question for a sharper one: “How do you plan dispatch today, and what breaks at peak?”
  5. Set the goal (3 minutes). She decides the call succeeds if the customer agrees to a site visit and a second call with the IT head.

The brief took 20 minutes instead of an hour, and the most valuable step was the check that removed an invented fact before it reached the customer.

Guardrails for AI in sales

AI in sales guardrails: what to avoid and what to do instead
Five guardrails that protect trust with buyers.
  • No invented facts. Work from pasted sources and ask AI to label assumptions.
  • Read everything before it is sent. Mass-produced “personalised” emails that nobody read damage your domain reputation and your brand.
  • Pricing and promises stay human. AI can explain your price list; only approved people can offer discounts or commit to delivery dates.
  • Respect calling and consent rules. AI doesn’t change your obligations. In India, commercial calls and messages fall under TRAI’s rules, including the DND preference register. In the UAE, telemarketing is regulated under Cabinet Resolution No. 56 of 2024, which requires TDRA approval and restricts calling hours, among other conditions. Check the current rules with your compliance team.
  • Protect customer data. Minimum data, approved tools, and nothing confidential in personal accounts.

Which workflow to start with

WorkflowAI doesThe rep or manager doesGood first step for
Call briefsSummarises sources, drafts questionsVerifies facts, sets the call goalEvery rep
Recaps and CRM notesStructures notes, drafts the emailChecks commitments, sends, savesTeams with patchy CRM data
Proposals and RFPsCompliance matrix, first draftsOwners approve every claimBid and pre-sales teams
Objection practicePlays the buyer, gives feedbackPractises, refines the playbookNew joiners
Pipeline reviewFlags stale or thin dealsCoaches, decides the forecastSales managers

Common mistakes to avoid

  • “Write me a cold email” with no context. Generic prompts produce generic emails that buyers recognise instantly. Give the AI the account research and one specific reason to reach out.
  • Personalisation theatre. Mentioning a prospect’s latest post in the first line isn’t personalisation if the rest of the email ignores their situation.
  • Letting AI do the thinking in discovery. The question bank prepares you; listening is still your job.
  • Skipping the manager. Teams adopt faster when managers use the same prompts in pipeline reviews and coaching.
  • No measurement. Track prep time, proposal turnaround and CRM completeness before and after. Our article on measuring the ROI of AI training shows how to keep the numbers credible.

A 30-day starter plan for a sales team

  1. Week 1: agree the tool, data rules and the “stays human” list; record baseline times for call prep, proposals and CRM updates.
  2. Week 2: practise account research and call briefs on five real accounts; collect the prompts that work in a shared library.
  3. Week 3: use call briefs, recaps and CRM notes on live deals; managers review a sample each week.
  4. Week 4: add proposals or objection practice, re-measure, and decide what becomes standard in your sales process.

Reps and managers rarely start at the same level, so an AI skills gap analysis helps decide who needs what. Once deals close, the handover matters too: our guides on AI in customer service and AI in accounting cover the teams that look after the customer and the invoice.

Questions people ask

The most practical uses are account research from real sources, account plans, one-page call briefs, discovery question banks, same-day recaps and follow-ups, proposal and RFP first drafts, objection handling practice, structured CRM notes and pipeline review preparation. AI prepares and drafts; reps verify facts, build the relationship and make every commitment.
It is changing the job rather than removing it. AI takes on research, drafting and admin, which gives reps more time for conversations, discovery and negotiation. Buyers still want to deal with a person who understands their business and can be held to a promise.
Start with the general assistant your company already licenses, such as ChatGPT, Microsoft Copilot, Claude or Gemini, and the AI features in your CRM. The workflows and review habits matter far more than the tool, and the prompts in this guide work across all of them.
Only into tools your company has approved, on work accounts, and only the data the task needs. Contact details are personal data under India’s DPDP Act and the UAE’s PDPL, and contracts and pricing are usually confidential. Never use a personal chatbot account for customer information.
Give the AI real context: your research on the account, the specific trigger for writing, what you know about the reader’s role and your own voice guidelines. Ask for short drafts, then edit them yourself. If you wouldn’t send it to the CEO of your best customer, rewrite it.
It can build a compliance matrix and first drafts from your approved content, which saves a great deal of time. Every claim about capability, delivery, security and pricing must be checked and approved by the person who owns it before submission.
To prepare pipeline reviews by flagging stale or thin deals, to build shared question banks and playbooks, to set up objection practice for new joiners and to coach from call recaps. Managers who use the same prompts as their reps make adoption much faster.
Record baseline times for call prep, proposal writing and CRM updates, plus proposal turnaround and CRM completeness, then compare after a month. Watch quality signals too, such as reply rates and how often managers have to correct proposals.

Sales is only as fast as the rest of the business

Proposals wait on pricing, contracts and delivery teams. If you want everyone who touches a deal using AI with the same care, our generative AI productivity programme has whole departments rebuild their own documents, reports and processes, and leave with a governed prompt library.
Generative AI for productivity
Talk to our team
Prefer email? Write to hello@skillopediagroup.com

About Skillopedia

Skillopedia is a corporate AI training firm, established in 2017, running hands-on programmes for teams across India, the GCC, Africa and Southeast Asia, with on-site and online sessions across the UAE. More about us