Sales has a small core (understanding the buyer's problem and moving the deal) wrapped in a large shell of prep, note-taking, follow-up, and admin. AI is very good at the shell. It can turn a 10-K and three press releases into a call brief, a recorded discovery call into CRM notes and a follow-up email, and a rough scope into a proposal draft.
The catch is that AI has no idea what your buyer said last Tuesday unless you tell it, and it will invent product specs, pricing, and contract terms if you let it. Treat everything it says about your offering as a draft to verify and everything it says about a prospect as a lead to confirm. The same goes for the law: CAN-SPAM, the TCPA, and state call-recording rules apply just as fully to AI-written sequences and AI-transcribed calls.
Quick wins this week
- Before your next discovery call, paste the prospect's About page, latest press release, and your contact's LinkedIn headline into a chat assistant and ask for five questions that prove you did your homework.
- Record a call with a consented notetaker like Otter or Fireflies, paste the transcript in, and ask for decision criteria, objections, next steps, and a 120-word follow-up in your voice.
- Give the AI a pipeline export with contact details stripped and ask which deals have gone quiet for more than 14 days and what a re-engagement note to each might say.
- Paste an objection word for word ('we already use [COMPETITOR]') and ask for three responses: curious, evidence-based, and a graceful exit. Practice them aloud.
What AI can do for sales professionals, task by task
Account and prospect research
Give a search-connected tool the company name, industry, and what you sell, and ask for a one-page brief: what the company does, recent news, likely priorities for your buyer's role, and three pain hypotheses to test on the call. Open the citations; executive names, 'recent' news, and headcounts are exactly where models go stale or invent.
Cold outreach that gets replies and stays legal
Provide your ideal customer profile, the trigger for reaching out, a real proof point, and your three best past emails, and ask for a short sequence with one ask per message. Then check the basics: accurate from-name and subject, your postal address and an opt-out path in every commercial email, and no texts to mobile numbers without documented consent. AI drafts words, not consent.
Call notes and CRM updates
Transcribe calls with a consented meeting assistant, then ask a chat assistant to extract MEDDIC or BANT fields, objections, competitors mentioned, and next steps in the exact format your CRM needs. Read the key moments of the transcript yourself before trusting a summary of a deal that matters; transcription errors on names, numbers, and dates are common.
Follow-up emails and recaps
Paste the transcript or your notes and ask for a follow-up under 150 words that restates the buyer's problem in their words, lists agreed next steps with owners and dates, and ends with one question. Check that every commitment is one you actually made; AI recaps sometimes upgrade 'we'll look into it' to 'we'll deliver it'.
Proposals and RFP responses
Load past winning proposals, the current pricing sheet, and the prospect's requirements into a Project or NotebookLM so the model draws only from your documents, then ask for a draft that maps each requirement to a capability and flags gaps honestly. Verify every price, SLA, and feature claim against the source; a hallucinated commitment in a signed proposal is a contract problem.
Objection handling and call practice
Ask the AI to role-play your toughest buyer persona through a discovery call, pushing on price, timing, and incumbents, then grade your answers. No real customer data is involved, so a consumer tool is fine, and it is the fastest way to stop being surprised on live calls.
Prompts for sales professionals
Replace the bracketed placeholders, paste into any chat assistant, and iterate on the result.
Pre-call brief
You are a sales research analyst. I sell [YOUR PRODUCT] to [BUYER ROLE] at [TYPE OF COMPANY]. I have a discovery call with [PROSPECT COMPANY] with [NAME AND TITLE]. Using the notes below and anything you can verify from public sources (cite them), write a one-page brief: 1) what the company does and how it makes money, 2) three recent events that might matter to my buyer, 3) three hypotheses about problems my product could solve, each phrased as a question I could ask, 4) likely objections. Flag anything you are unsure about instead of guessing. [PASTE COMPANY INFO OR NOTES]
Tip: Run it in a search-connected tool and open every citation before the call.
Cold email sequence that respects CAN-SPAM
You are a B2B outbound writer. Write a [NUMBER]-touch email sequence to [BUYER ROLE] at [COMPANY TYPE] who [TRIGGER, E.G. JUST POSTED A JOB FOR A SECOND SCHEDULER]. My offer: [ONE-SENTENCE VALUE PROPOSITION]. Proof point: [REAL PROOF POINT]. Rules: each email under 90 words, plain text, no fake familiarity, a subject line that honestly describes the body, one question as the ask, and a placeholder for our postal address and an opt-out line in the signature. Match the tone and length of these past emails that got replies: [PASTE THREE EMAILS]
Tip: If your sequence tool adds the footer automatically, delete the placeholder, but confirm it does.
Call transcript to CRM fields
You are my sales operations assistant. Below is a transcript of a discovery call with [PROSPECT COMPANY]. Fill in this template, writing 'not discussed' where needed: pain points in the buyer's words, current solution, decision criteria, decision process and who else is involved, timeline and budget signals, competitors mentioned, risks to the deal, agreed next steps with owner and date. Then draft a follow-up email under 150 words that restates their problem, confirms next steps, and asks one question. Do not invent commitments that were not made on the call. [PASTE TRANSCRIPT]
Tip: Match the template field names to your CRM so the paste-in takes seconds.
Objection role-play
Act as [BUYER ROLE] at a [COMPANY SIZE] [INDUSTRY] company who is skeptical about buying [PRODUCT CATEGORY]. You currently use [INCUMBENT OR STATUS QUO] and worry most about [TOP CONCERN, E.G. IMPLEMENTATION TIME]. Run a realistic discovery call with me: ask hard questions, push back on price and timing, and do not make it easy. Stay in character until I type 'debrief'. Then grade my responses on listening, clarity, and objection handling, and give me three specific things to say differently next time.
Tip: Do it in voice mode if your tool has one; typing hides bad habits.
Proposal draft from your own documents
You are a proposal writer for [COMPANY]. Using only the documents I have provided ([LIST DOCUMENTS, E.G. PRICING SHEET, PAST PROPOSALS, PRODUCT SPECS]) and the requirements below, draft a proposal for [PROSPECT COMPANY]. For each requirement, state how we meet it and cite the supporting document; if nothing supports it, write 'GAP: confirm with product team' instead of guessing. Include an executive summary in the prospect's own words about their problem, a scope section, pricing taken exactly from the pricing sheet, and assumptions. Do not invent SLAs, timelines, or discounts. [PASTE REQUIREMENTS]
Tip: The GAP list is your to-do list before the proposal leaves the building.
Want a prompt for something else? Use the Prompt Builder or browse career and job search prompts, customer communication prompts, meetings and summaries prompts, sales and outreach prompts, writing and email prompts.
Skills to build
Working from the transcript, not from memory
Why: Your notes are a filtered version of the call; the transcript is what the buyer actually said, in words that persuade when you reuse them.
How: Standardize on one consented notetaker for every call and one extraction prompt. After two weeks, compare your CRM notes before and after.
Research verification
Why: One wrong fact in a brief (a merged company, a departed executive) costs you credibility in the first two minutes of a call.
How: Use search-connected tools only and open every citation. Learn what models get wrong most: dates, headcounts, and who reports to whom.
Building a personal prompt library
Why: The difference between generic outreach and a reply is specificity: trigger, proof, ask. The same is true of the prompt.
How: Save a pre-call brief, sequence, recap, and objection-drill prompt in a Project. Refine each after every use.
Compliance basics for outreach
Why: You are the person who hits send. Fines and blocked domains land on the company; the reputation damage lands on you.
How: Learn the CAN-SPAM checklist, the TCPA consent rule for texts and calls, your state's call-recording rule, and your sequence tool's opt-out handling. Have legal or RevOps review your templates yearly.
Clean pipeline data
Why: AI can only analyze a pipeline with clean stages, next steps, and close dates. Garbage in, confident nonsense out.
How: Spend an hour a week on CRM hygiene with AI help: paste the stalled-deals list (contacts stripped) and ask for a re-engagement plan for each.
Tools worth knowing
ChatGPT
The general-purpose AI assistant most of your coworkers already use.
Claude
A careful writing and analysis assistant that shines on long documents.
Perplexity
An answer engine that cites its sources, built for research rather than chat.
Otter.ai
Meeting transcription and summaries that follow you into Zoom, Teams, and Meet.
Fireflies.ai
A meeting note-taker with deep integrations, search across calls, and an AI assistant.
Microsoft Copilot
AI inside Word, Excel, Outlook, and Teams, with your company's data protections.
Cautions for sales professionals
Commercial email must have accurate headers, a truthful subject line, your physical postal address, and an opt-out honored within 10 business days, for one AI-written note as much as for a 5,000-send sequence. Texts and calls to mobile numbers generally need prior express consent under the TCPA (written consent for marketing), and the FCC has confirmed that AI-generated voices count as artificial voices under that law. Never let an automation tool text or call a list without documented consent, and check Do Not Call rules with your compliance team.
Meeting notetakers record, and several states, including California, require all parties on a call to consent. Announce the recorder at the top of every call, get a verbal yes, and stop recording if anyone objects. Store transcripts where your company controls them.
AI will state a feature, integration, price, or SLA you do not have, in perfectly confident prose. Anything that touches what you promise a customer, in an email, a proposal, or a deck, gets verified against a source document first. A hallucinated commitment a buyer relies on is a legal problem, not a typo.
Prospect names, emails, phone numbers, and deal notes are personal data and often confidential under NDAs. Never paste confidential data into a consumer AI tool unless your organization has approved it: use the approved tool with training disabled, strip contact details from exports, and check the NDA before uploading a customer's documents.
LinkedIn and most email providers prohibit or throttle automated messaging and scraping, and AI-driven bulk outreach is the fastest route to a restricted account or a blacklisted sending domain. Use AI to write better messages that you send through approved channels, not to send more than a human could.
Your 30-day plan
- Week 1: Set up a consented call recorder and a chat assistant with a saved pre-call brief prompt, and use both on every external call this month. Rewrite your top outreach template with AI and run it through the CAN-SPAM checklist before sending.
- Week 2: Standardize call-to-CRM extraction with one prompt and one template that matches your CRM fields. Time your post-call admin before and after.
- Week 2: Run three objection role-plays on your toughest recurring pushback and write down the responses that worked.
- Week 3: Build a Project or notebook with your pricing sheet, past proposals, and specs, and draft your next proposal from it. Clean your pipeline with AI help: stalled deals, missing next steps, a re-engagement note for each.
- Week 4: Review reply rates, admin time, and note quality with your manager. Keep the workflows that moved a number and share the prompts with the team.
Frequently asked questions
Will AI replace salespeople?
Can I use AI to write cold emails?
Is it legal to record sales calls with an AI notetaker?
Which AI tool is best for sales?
Terms used on this page
Related roles
- AI for MarketersDraft campaigns faster, mine customer feedback for messaging, and turn reporting into a ten-minute job, without drifting off-brand or into a CAN-SPAM problem.
- AI for Customer Service RepsLet AI draft the reply, find the policy, and summarize the thread so you can spend your energy on the customer in front of you. Here is what to hand off, what to check, and what never goes into a chat tool.
- AI for Real Estate AgentsAI writes the listing description, summarizes the 60-page inspection report by severity, drafts the follow-up sequence, and turns your MLS stats into a neighborhood update. Fair housing, the contract, and the client are still on you.
- AI for Insurance AgentsAI explains replacement cost versus actual cash value to a client in two paragraphs, drafts the rate-increase email you dread, and builds a quote comparison from three dec pages. The advice, the license, and the E&O exposure stay yours.