9 AI Prompts to Improve Sales Productivity: A Rep Guide

AI improves sales productivity fastest when it removes the rep's daily non-selling work: planning the day, preparing for calls, researching accounts, writing follow-ups, updating the CRM and restarting stalled deals. The nine copy-ready prompts below each take one of those jobs off a rep's plate, with what good output looks like and the mistake that wastes the time you saved.

Key takeaways

• Reps spend most of their week on work that is not selling, so the biggest productivity gains come from admin, prep and follow-up.

• Every prompt works better with real input: your deal list, the call transcript, the buyer's emails.

• Ask for short outputs with a word limit and one goal. Long AI output just moves the reading work around.

• Keep the rep in control: AI drafts the email and the CRM update, the rep approves.

• In Proshort, Ask Aria answers the first two prompts from your live deals and calendar, and can send them every morning.

Why do sales reps need AI prompts for productivity?

Because selling is a minority of the job. Salesforce's State of Sales report (7th edition), based on a survey of 4,050 sales professionals in August and September 2025, found reps spend 40% of an average workweek selling and 60% on nonselling work such as data entry and prospecting. Every prompt in this guide targets part of that 60%.

The prompts follow two rules from the model makers. Anthropic advises treating the model like a brilliant but new employee who lacks context, so each prompt says exactly what to produce and how long. OpenAI's prompt engineering guide recommends giving the model reference text it would not otherwise have, so each prompt asks you to paste or connect the real deal data.

The 9 prompts at a glance

Prompts for before the call

1. Plan the day around the deals that matter

When to use it: First thing every morning, before email.

Here are my open deals with stage, amount, close date, last buyer activity and next step, plus today's calendar: [paste]. Which five deals need my attention today and why? For each, give the one action that moves it (a call, an email, a question to answer) and estimate how long it takes. Flag any deal with no buyer activity in [14] days.

What good output looks like: Five deals, one action each with a time estimate, and a short list of deals going quiet. It should fit on a phone screen.

Common mistake: Asking for a to-do list without deal data. The model will invent generic tasks such as "follow up with prospects".

2. Brief me before every call

When to use it: Fifteen minutes before each external meeting.

Brief me for my [time] call with [company]. Use the notes, emails and past call summaries below: [paste]. Give me: who is attending and what each person cares about, what we agreed last time, open questions or objections, the one thing I must learn on this call, and a suggested opening question. Keep it under 150 words.

What good output looks like: A short brief with names, the open objection and one goal for the call, not a company history lesson.

Common mistake: Asking for "everything about the account". Long briefs get skimmed; set a word limit and a single goal.

3. Research an account in five minutes

When to use it: Before first outreach or a first meeting with a new account.

Using the public information and our CRM history below, summarize [company] for a first sales conversation: what they sell, recent changes (leadership, funding, launches, hiring), likely priorities for a [persona], and two hypotheses about why they might need [our category]. Mark anything you are unsure of as unverified. [paste sources]

What good output looks like: Two testable hypotheses and a clearly marked list of unverified facts, so the rep checks before quoting them.

Common mistake: Trusting the model's memory of the company. Paste current sources; training data goes stale.

4. Prepare discovery questions for one persona

When to use it: Before a discovery call with a role you rarely sell to.

I am meeting a [title] at a [industry] company about [problem]. Write eight discovery questions in the order I should ask them: two about their current process, two about impact in numbers, two about how they decide, and two about timing. After each, add the follow-up I should ask if the answer is vague.

What good output looks like: Open questions grouped by purpose, each with a follow-up for vague answers. The follow-ups are the valuable part.

Common mistake: Accepting yes/no questions. If an answer can be "yes", rewrite the question.

Prompts for after the call

5. Write the follow-up email from the call

When to use it: Within an hour of every customer call.

Using the call transcript below, draft a follow-up email to [buyer]. Use the buyer's own words for their problem, list the action items with owners and dates, confirm the next meeting, and keep it under 150 words. Do not add anything that was not said on the call. [paste transcript]

What good output looks like: A short email the buyer could forward internally, written in their language, with owners and dates.

Common mistake: Letting the email promise things nobody agreed to. Always read before sending; nothing should go out without the rep.

6. Update the CRM from the call

When to use it: Straight after the call, alongside the follow-up email.

From this transcript, fill in these CRM fields: next step (with date and owner), close date (with the buyer statement that supports it), metrics, economic buyer, decision criteria, decision process, pain, champion and competitors mentioned. For each field, quote the line from the call that supports it, or write "not discussed". [paste transcript]

What good output looks like: Every field either backed by a quote or marked "not discussed". The gaps tell you what to ask next time.

Common mistake: Letting the model fill gaps with guesses. "Not discussed" is a useful answer; an invented champion is not.

7. Prepare for the objection you will hear next

When to use it: The day before a follow-up call with an open concern.

In the last two calls with [company], what concerns did the buyer raise, in their exact words, and which were left unresolved? For each unresolved one, give me a short answer, one proof point from [our case studies or docs], and a question that checks whether my answer landed. [paste transcripts]

What good output looks like: Unresolved concerns in the buyer's words, each with an answer, a proof point and a check-in question.

Common mistake: Preparing for generic objections. The objection you will hear is usually the one this buyer already raised.

8. Restart a stalled deal

When to use it: When a deal has had no buyer reply for two or more weeks.

This deal has gone quiet for [N] days. Here is the history: [paste last call summary and emails]. Write a three-sentence email that references something specific the buyer said, offers one useful thing (an answer, a resource or a shorter path), and asks a question that is easy to reply to. Then suggest one other person at the account I could reach instead.

What good output looks like: A short, specific note that gives before it asks, plus a second contact to multi-thread.

Common mistake: Writing "just checking in". It signals you have nothing new; give the buyer a reason to reply.

9. Run a Friday self-review

When to use it: End of every week, 15 minutes.

Here is my calendar and activity for this week, plus summaries of my customer calls: [paste]. How many hours went to customer conversations, prep, internal meetings and admin? Which two internal meetings or tasks could I drop or shorten next week? On my calls, what share of the time did I talk, and where did I miss a chance to ask a follow-up question?

What good output looks like: A time split, two things to cut and one coaching point from your own calls.

Common mistake: Reviewing only activity counts. More calls is not the goal; more time in useful buyer conversations is.

What should you look for in a sales productivity tool?

Prompts in a chat window still need copying and pasting. If you are choosing a sales productivity platform to run this work for you, check five things:

• It works where reps already work: CRM, calendar, meeting tool, email and Slack, with no new tab to remember.

• It helps during the call, not only after: the most expensive minute is the one where the buyer asks a question the rep cannot answer.

• Answers come with sources: reps and managers can click through to the call or record behind every claim.

• Nothing goes out without the rep: emails and CRM updates are drafted, then approved.

• Security is certified: look for SOC 2 Type 2 and ISO 27001 before connecting calls and CRM data.

For a side-by-side comparison of vendors, see our guide to sales productivity tools with data-driven insights.

How does Proshort run these prompts for you?

Ask Aria sits on top of every deal, call and CRM record. "Which deals need my attention today?" and "Brief me on all my upcoming calls today" are the kind of plain-English questions it answers, with sources, and you can schedule them as a daily 8:00 email so prompts 1 and 2 are waiting before the first call. Proshort's app in ChatGPT and its Claude connector (MCP) let you ask the same questions from the assistant you already use.

After the call, AI follow-up emails draft the recap in the buyer's own words with action items, owners and the next step, and task automation handles tasks and CRM updates after the rep's OK. During the call, Aria Live puts the answer, proof point and next question on the rep's screen in under a second. Only the rep sees it.

"For me, this becomes a very overarching sales AI use case where it is not about call recording. It is about using AI to manage and structure my mailbox, my communications tied to a particular deal." Amit Tomer, VP of Strategic Sales, Aviz Networks (read the Aviz Networks case study)

Managers can take the next step with 7 AI prompts for pipeline reviews, and leaders with 12 AI prompts for sales strategy.

Frequently asked questions

What is the best AI prompt for sales productivity?

The morning prioritization prompt (prompt 1) usually saves the most time, because it decides where every other hour goes. Feed it your open deals and calendar and ask for five deals with one action each.

Which AI assistant tools improve sales productivity?

General assistants such as ChatGPT and Claude handle drafting and research when you paste in context. Sales platforms connected to your calls, CRM and calendar remove the copy-paste step and can act before, during and after the call. Many teams use both.

How much time can AI save a sales rep?

It depends on how much of the week goes to admin today. Measure it yourself: run the Friday self-review prompt for two weeks before and after adopting AI follow-ups and CRM updates, and compare the hours spent on customer conversations.

Is it safe to paste call transcripts into ChatGPT?

Use a company-approved enterprise workspace that does not train on your data, and follow your security policy on customer information. A connected platform with SOC 2 Type 2 and ISO 27001 certification, such as Proshort, avoids pasting data by hand.

Should AI update the CRM automatically?

AI should draft the update and show the call evidence behind each field, and the rep should approve it. That keeps the CRM accurate without turning reps into data-entry clerks, and it stops guesses from becoming forecast inputs.

See prompts 1, 2, 5 and 6 running on your own calendar and deals: book a Proshort demo.