35 AI agent prompts for Sales
Copy-paste, beginner-friendly prompts to help you research accounts, qualify leads, write outreach and keep your CRM clean. Each one is based on a real-world example and works with ChatGPT, Claude or an AI agent. 28 are free below, and 7 more are in AgentPlaybook Pro.
Free Sales prompts
Lead qualifier
Score inbound leads against your ideal customer profile.
My ideal customer: [industry, size, budget, problem]. Score each lead 1-10 for fit, explain in one line, and suggest the next action (call, nurture email, discard). Leads: [paste]
Account research brief
Prepare a 1-page brief before a sales call.
Research [company name + website]. Give me: what they do, recent news, likely priorities, decision makers' roles, 3 pain points my product [describe] could solve, and 3 smart questions to ask on the call. Cite your sources.
Call notes to follow-up email
Turn sales-call notes into a follow-up and CRM update.
From these call notes [paste], write: (1) a follow-up email recapping their goals, agreed next steps and dates; (2) a CRM note with deal stage, budget, timeline, objections.
Duplicate contact finder
Clean your CRM by spotting duplicates and suggesting merges.
Here is a CRM export (name, email, company, phone): [paste CSV]. Find likely duplicates (similar names, same domain, same phone), list them in pairs with a confidence level, and suggest which record to keep.
Lead enrichment + draft reply
When a new lead arrives, enrich it and draft a reply you send in two clicks.
Build me a step-by-step automation plan: when a new form lead arrives, look up their company and role, score fit vs [ICP], add the info to my CRM [tool], and draft a personalized follow-up email for my approval. List each step and the tool to use.
Stale quote chaser
Surface sent quotes that haven't been answered before they go stale and draft a nudge for each, like the Quote Follow-Up capability Prokeep launched for distributor counter teams.
Here is my list of quotes I've sent: [paste: customer, quote #, items, amount, date sent, last contact, status]. Flag every quote with no reply after [X] days, plus any that expire within [Y] days. Rank the flagged ones by value and how likely they are to close. For each, draft a short, friendly follow-up in my voice that mentions the specific items and offers an easy next step (confirm stock, adjust quantities, or a quick call). Don't send anything; give me the list to review.
Daily client activity digest
Each morning, summarize what every client did across messages, saved searches and price alerts, with suggested next steps and drafted messages you approve, like Realtor.com's RealAssist AI Daily Digest for real estate agents.
Here's yesterday's activity for my clients: [paste: client name, messages, what they viewed, saved or searched, alerts such as price changes, my notes]. For each client, give me a 2-line summary of what changed, how ready to act they seem (hot / warm / cold) and why, and one suggested next step. For the top [3] clients, draft a short message in my tone (here's a sample of how I write: [paste]) that I can review before sending. Put the clients who need action today at the top.
Find local businesses already buying ads
Build a prospect list of businesses already advertising in your area, check that they actually serve your market and collect their published contact details, like the weekend agent John Apolinar set up with Hermes and ChatGPT/Codex for his local media company.
Help me build a prospect list for selling [ad space / my service] in [city or region]. Find businesses that are already advertising locally (for example in local papers, radio, billboards, local social ads, or sponsorships) in these categories: [list]. For each one, check whether they actually serve [my area] and note how you confirmed it. Collect only contact details they publish themselves (website contact page, listed phone or email, booking page), plus the source link. Put it in a table: business, category, where they advertise, serves my area (yes/no/unsure), published contact, source. Skip anyone you can't confirm, and don't contact anyone.
Turn new newsletter subscribers into qualified pitches
Catch each new newsletter subscriber, research them against your ideal customer profile, and draft a personal pitch only for the ones you approve, like Sharyph's no-code Substack Subscriber Pitch Agent with one human approval checkpoint before anything sends.
You help me find warm prospects among my newsletter subscribers. My offer: [describe what you sell and the price]. My ideal customer: [e.g. founders with a live product, recently raised, B2B SaaS]. Here are this week's new subscribers (name, email, profile link): [paste from your Substack/beehiiv notification emails or export]. For each one: 1) look at their public profile, website, X and LinkedIn; 2) decide Fit / Maybe / Not a fit against my ideal customer, with the evidence. If you can't verify a signal, don't count it in their favor. 3) For Fit only, draft a short, personal first email in my voice ([paste 2-3 emails I've written]) that refers to something specific about them, plus up to [3] polite follow-ups. Put everything in a table. Don't send anything: I'll mark which drafts are approved, and follow-ups must stop as soon as someone replies.
Ask your CRM agent exact pipeline and win-rate questions
Treat your CRM as something agents can query directly for exact answers on win rates, weighted pipeline, and risky accounts, like Attio's October 2026 hybrid architecture where Ask Attio and SQL let teams answer questions such as which lead sources win deals over $50,000 without exporting data.
You have read access to my CRM ([Attio / Salesforce / HubSpot]). Answer this question with exact numbers and the logic you used: [e.g. which lead sources have the highest win rate on deals over $50,000 / weighted pipeline by region and owner / open opportunities tied to churned customers]. Show a short table of results, name the filters and date range you assumed ([last 12 months] unless I say otherwise), and list any records you excluded and why. Don't create, update, or delete any CRM records, and don't email anyone. If you need a SQL-style query, show it so I can inspect it.
Guide a car buyer from call to test drive
Run a multi-step buying conversation that asks about budget and needs, recommends cars, books a test drive, and follows up on financing—like Cars24's OpenAI voice and chat agents that handle 1M+ conversation minutes a month across buying, selling, and after-sales.
You are a car-buying assistant for [dealership / marketplace]. A customer just contacted us on [voice / chat / WhatsApp]. Ask about budget, family size, commute, and preferred body type (max 4 questions), then recommend up to [3] cars from this catalog: [paste or link]. For the best fit, propose a test-drive slot from [availability], draft a confirmation message, and list financing docs we still need. After the visit, draft a follow-up that offers another car or next step. Catalog and policy: [paste]. Don't invent stock, prices, or APR you don't have. Don't book or charge anything until I confirm.
Re-engage sellers who went quiet after 10 days
Bring dropped vehicle-seller leads back into the funnel with a qualify-and-rebook agent, like Cars24's AI re-engagement that recovered 12% of previously lost seller leads after about 10 days of silence.
Here are seller leads that went quiet for [10+] days: [paste name, vehicle, last touch, reason if known]. For each lead, draft a short re-engagement message that: 1) acknowledges the gap, 2) asks one question to check if they still want to sell, 3) offers a concrete next step (inspection slot or updated price range), and 4) stays under [80] words. Flag leads that look like a hard no or need a human. Tone: [friendly / local market]. Don't send messages, don't change CRM stages, and don't promise a buy price you can't support. Wait for me to approve which leads to contact.
Draft an RFP response from past wins
Pull language from proposals you've already won—plus product and pricing details—into a grounded RFP draft for human review, like NFP's Microsoft Copilot Cowork agents that execute document-heavy insurance work instead of only drafting.
You are my proposal assistant. Past winning proposals and product/pricing notes: [paste or folder]. New RFP / RFI: [paste]. Draft a response that reuses only facts and language grounded in the past wins and notes. Structure: cover letter, requirement-by-requirement answers, pricing caveats, and a risk/open-items list. Cite which past proposal each major claim comes from. Flag any RFP requirement we have never answered before. Don't invent certifications, SLAs, or prices. Don't email the buyer or change CRM stage—leave everything as DRAFT until I approve.
Match a messy customer order to your product catalog
Paste a customer's texted, emailed, or photographed order and get it matched to your exact products, reading the whole order for context instead of one line at a time, ready for a rep to check. Prokeep launched this as the Match Agent in its Order Pro agents for wholesale distributor counter teams.
You help our counter team turn customer requests into orders. Our catalog: [paste or attach product list with SKU, name, size, and unit]. This customer usually buys: [paste recent orders, if any]. Here is their request: [paste the text, email, or a transcription of the photo or handwritten list]. Read the whole request first to work out the job or project it's for, then for each line give: what they asked for, the SKU you'd match, quantity and unit, and how confident you are (high, medium, or low). For anything low-confidence or missing from the catalog, suggest the closest substitute and write one short question I can text back to confirm. End with a draft reply to the customer that I'll review before sending.
Keep CRM and prospect records continuously fresh
Describe the accounts you care about and let a fleet of research agents keep company, contact, and buying-signal fields updated from many sources—instead of SDRs manually cleaning the CRM. Floqer's customers report cutting prospecting time by as much as 70%.
You are my CRM hygiene agent. Ideal customer profile: [industry, size, geography, tech stack, buying trigger]. Target list or segment: [paste account names or CRM export]. Fields we must keep current: [e.g. company size, funding, key contacts, emails, recent news]. For each account, return: confirmed company facts with sources, best-fit contacts with role and confidence, any job-change or funding signal from the last 90 days, and a one-line next action for the owner. Flag stale or conflicting CRM values we should overwrite only after a human check. Skip anything you cannot verify from a real public source.
Answer every catering or event inquiry fast, even when you're not there
Restaurants and venues lose catering and private-event bookings when inquiries sit until the owner is back in the building. One restaurant owner shared on Oct 21, 2026 that putting written checklists, event runbooks and an AI agent on every inquiry took catering from 1 or 2 bookings a month to 7 to 12.
You answer catering and private-event inquiries for [restaurant name]. Our offer: [paste menus/packages, prices per person, minimum headcount, deposit, service area, lead time, dates we're already booked]. For each new inquiry below: Step 1: pull out the date, time, headcount, location, budget, event type and any dietary needs; list what's missing. Step 2: write a warm reply within our rules that suggests 2 packages that fit, gives a price range, and asks only for the missing details. Step 3: if the date is already booked, the headcount is under our minimum, or they ask for a custom discount, don't promise anything; flag it for me instead. Step 4: add a one-line summary I can paste into our event tracker, and draft a polite follow-up to send in 2 days if they don't reply. Never confirm a booking or take a deposit; that's mine to approve. Inquiry: [paste]
Get a weekly list of companies that just changed owners, leaders or funding, with a tailored pitch for each
Companies are most open to a new supplier or partner right after something changes: a new owner, a new leader, a big new project, an acquisition or a funding round. Have an AI agent scan for these moments every week, write a short brief on each company, sketch what you could do for them and draft a personal first message, with you approving every send. Applied technology studio CapePoint posted on Oct 22, 2026 that it built exactly this sales agent and ran it on its own pipeline first, so first calls are spent on what's actually on the prospect's list.
Act as my weekly prospecting researcher. I sell [what you sell] to [type of company, size, region]. Every [Monday], search news, press releases, LinkedIn and funding announcements from the past 7 days for companies matching that profile that had one of these changes: new owner, new CEO or department head, a new project or location, an acquisition, or a funding round. For each company (max [10]): 1) a 5-line brief: what they do, what changed, date and source link; 2) why that change could create a need for what I sell; 3) 2-3 specific ideas I could offer them; 4) a short, friendly first email or LinkedIn message that mentions the change naturally and asks one question. Put it all in a table and skip any company where you can't find a real source. Don't send anything; I'll approve each message.
Read incoming quote requests by email, in any language, and draft a priced quote with the best route options
Freight forwarders, couriers and other shipping businesses get a steady stream of loosely written quote requests by email, often in several languages, and someone has to work out the route, service level and price for each one. An AI agent can sort the requests, pull out the shipment details, compare route options and draft a full quote for a person to check, while unusual or high-value requests go straight to the sales team. Arab News reported on Oct 22, 2026 that Saudia Cargo's AI workers now classify multilingual email rate requests, calculate multi-leg options, pick service tiers such as pharma or fresh cargo, and prepare personalized quotes within seconds, with every action logged and complex cases escalated to people.
I run [business name], a [freight forwarder / courier / trucking / shipping] business. Here are today's quote request emails: [paste emails]. Here is my rate sheet and rules: [paste rates, surcharges, minimums, service levels such as express, temperature-controlled or fragile, and routes we offer]. For each email: 1) Translate it to English if needed and say which language it was in. 2) Label it: real quote request, follow-up, or not a quote. 3) Pull out origin, destination, dates, weight, size, number of pieces, goods type and any special handling. List anything missing. 4) Give up to 3 route or service options with price, transit time and why you picked each, using only my rate sheet. 5) Draft a short, friendly reply in the customer's language with the recommended option and the questions needed to confirm. Flag for me instead of quoting if the shipment is over [value/weight], needs dangerous-goods handling, or doesn't fit my rate sheet. Don't send anything; I'll approve each quote.
Text back every new lead within minutes, day or night, and book the appointment before they call someone else
Home service businesses lose jobs when a lead from an ad or mailer sits unanswered, because the customer just moves on to the next company. An AI agent can text every new inquiry right away, answer basic questions, and book the visit, handing off only the tricky ones. ServiceTitan reported that True Pros Heating and Air in Layton, Utah, which runs its office with one full-time customer service rep, used an AI text agent called Sam that lifted its inbound lead conversion from about 9.2% to 28.5%, booked 51 jobs on its own across 185 conversations, and cut escalations from 22.7% to 0.6%. In one example a customer messaged at 8:33 p.m., got a reply at 8:35 and was booked by 8:36.
You are the text-message booking assistant for [business name], a [HVAC / plumbing / electrical / cleaning / other] company serving [area]. New lead details: [paste name, phone, what they asked for, where the lead came from]. Our services and prices: [paste]. Available appointment slots: [paste or say 'check my calendar']. 1) Write a friendly first text (under 300 characters) that greets them by name, confirms what they need and offers two specific time slots. 2) For each reply I paste back, answer their question using only the info above, and keep steering toward a booked time. 3) Once they pick a time, confirm the address, the problem in their words and the best callback number, then write the final confirmation text. 4) Hand off to me instead of answering if they mention an emergency like gas smell, flooding or no heat in freezing weather, ask for a price you don't have, are upset, or want to cancel. Keep the tone warm and local, never pushy, and never promise anything not listed.
Get a morning report of every lead nobody called back, every overdue follow-up and every appointment missing from your CRM
Leads rarely go cold because nobody cares; they go cold because a callback slipped or a new form fill sat unseen for a day. An AI agent can check your CRM every morning, list the leads with no first contact, the follow-ups that are past due and the appointments that were never logged, and send that list to the person who fixes it in the chat app your team already uses. Ryan Murray, who owns a 21-person shutter and shade manufacturer and installer, shared on X on Oct 17, 2026 that his team built this check on their GoHighLevel CRM with Claude: one report caught a lead with no reply for 19 days and a promised callback more than a week late, and first-hour lead response misses fell from 9 to 1 in a week (he says not all of that is down to AI).
You are my sales follow-up checker. Here is a CRM export of open leads: [paste or connect: lead name, source, date created, owner, last contact date, next follow-up date, notes, any appointment booked]. Today is [date]. Our rules: first contact within [1 hour / same business day]; follow up every [2] business days until they reply, up to [4] tries; every booked appointment must be on the calendar and logged in the CRM. Do this: 1) List leads with no first contact yet, oldest first, with how long they've waited. 2) List follow-ups that are past due, with how many days late and what was promised (quote the note). 3) List appointments mentioned in notes or the calendar that aren't recorded properly in the CRM. 4) Group everything by owner so each person sees only their own list, with one suggested next action per lead. 5) End with three numbers: leads waiting for first contact, overdue follow-ups, and first-hour misses yesterday. Keep it short enough to read in a team chat. Don't contact any lead or change any record; this is a checklist for my team.
Turn a rental or property enquiry into a booked viewing: an AI clarifies what the client wants, shortlists only places that fit, answers objections and agrees times with the owner
Getting listings online is the easy part for a property agency; the time goes into each client: working out what they really need, picking the few places that fit, handling 'too far' or 'too pricey', and going back and forth with owners on dates, price and terms before anyone sees a property. An AI agent can run that whole back-and-forth in chat and hand your team a confirmed viewing, so a person only does the showing. Nick Irbitskii, director of Unicorn Property in Bali, posted on X (Oct 22, 2026) that their AI agent now clarifies each request (budget, dates, area, must-haves), builds a shortlist of only the villas that fit, handles objections, talks to the owner about check-in/check-out, price and terms, and books every viewing, while a human just shows the villa.
You are my property enquiry assistant for [agency name], which rents/sells [type of property] in [area]. Here are our current listings: [paste or attach: name/ID, location, price, available dates, bedrooms, key features, owner contact rules]. Here is a new client enquiry: [paste the message or chat]. Do this: 1) List what the client has told us and what's still missing (budget, dates or move-in, area, number of people, must-haves, deal-breakers), and write a short, friendly reply asking only for the missing pieces. 2) Once I paste their answers, shortlist the 3-5 listings that genuinely fit, with one line each on why, and say which ones you left out and why. 3) For likely objections ('too far', 'too expensive', 'need an earlier move-in'), draft an honest answer for each using only our listing facts, plus a closer or cheaper alternative if we have one. 4) Draft a short message to each owner confirming availability for the client's dates, the price and the key terms, and proposing two viewing times. 5) When I paste the owners' replies, draft the viewing confirmation for the client (address, time, who meets them) and a one-line summary for my team. Never promise a price, discount or term the owner hasn't confirmed, and don't send anything yourself: give me drafts to approve.Spot companies about to hire before they post the job: have an AI watch filings and announcements for hiring signals and draft outreach for you to approve
Recruiters, agencies and B2B service firms win more work when they reach a company the moment it expands, not weeks later when the job ad is live and every competitor is calling. An AI agent can watch public company news, regulatory filings and job boards for signs a business is growing in your niche, match each signal to people or services you already have, and draft a short outreach note for you to check. Emergent reported (Oct 23, 2026) that Sydney recruiter Alicia Keenan rebuilt Aicila Group around four AI agents, including one that monitors regulatory filings and company announcements for hiring signals such as a European animal-health company expanding into Indonesia or Vietnam, while she reviews and approves every shortlist and message before anything is sent.
You are my business development scout. My business: [what you do, e.g. recruitment for animal-health and pharma roles / bookkeeping for clinics]. My target market: [industries, countries, company sizes]. My existing network or offer: [paste a short list of candidate profiles, case studies or services]. Do this: 1) List the public signals that a company in my market is about to need me (for example new site or country expansion, funding, new product approvals, leadership changes, a jump in job posts) and where to check each one for free. 2) From the news, filings or job posts I paste here [paste items], pick the ones that are real buying or hiring signals and explain why in one line each, ranked by how urgent they are. 3) For the top 5, match each to the people or services from my list that fit best. 4) Draft a short, specific outreach email for each (under 120 words, referencing the signal, no hype) for me to review. 5) Give me a simple weekly routine to repeat this. Never contact anyone yourself; I approve and send every message.
Reply to a new buyer lead within minutes: a first text, three qualifying questions and a booked call, without sounding like a bot
Portal and website leads go cold fast, and most agents reply hours later with a generic template. This prompt drafts a short first text that references the exact listing the lead asked about, asks no more than three easy qualifying questions, and proposes two concrete call or showing times, plus the follow-ups if they don't answer.
A new buyer lead just came in. Lead details: [name, phone/email, source (Zillow, Realtor.com, my website, open house), the listing or search they asked about, their message]. My details: [name, brokerage, area I cover]. My availability for calls or showings in the next 48 hours: [list]. Write: 1) a first text under 300 characters that uses their name, mentions the specific property or search, answers their question if I gave you the answer, and offers two specific times; 2) three qualifying questions I can ask over the next messages (timeline, pre-approval, must-haves), one at a time, phrased casually; 3) follow-up texts for day 1, day 3 and day 7 if there is no reply, each different and under 200 characters; 4) a 2-line CRM note. Don't promise availability, prices or anything about the property that isn't in the details above. Include an opt-out line in the first text if I'm texting someone for the first time: [yes/no].
Follow up with every open-house visitor the same evening, with a different message for buyers, neighbours and visitors who already have an agent
Open-house sign-in sheets often sit in a folder for days. This prompt sorts the sign-ins by type and drafts a personal same-day follow-up for each group, using what you noted about each visitor, plus a short summary you can send the seller.
Here are today's open-house sign-ins for [address]: [paste: name, email/phone, has an agent? (yes/no), notes I took, e.g. 'loved the kitchen, worried about street noise', 'neighbour, thinking of selling next spring']. Property highlights and price: [list]. Sort visitors into: active buyers without an agent, buyers with an agent, neighbours, and other. For each person, draft a short follow-up (email or text, whichever contact I have) that mentions something specific from my notes. For buyers without an agent, offer one useful next step (disclosures, a second showing, similar homes). For buyers with an agent, keep it to a thank-you and say I'm happy to send details to their agent; don't solicit them. For neighbours, offer a free valuation chat only if my notes say they're curious. Then write a 5-line summary for the seller: number of groups, common positives, common objections (in their words), and any likely offers. Don't invent feedback that isn't in my notes.
Explain your comparable sales to a seller in plain English before the listing appointment
Sellers often fixate on one high sale or an online estimate. Paste the comparables you pulled and your adjustments, and the prompt writes a clear one-page explanation of the suggested price range, why each comp matters, and answers to the objections sellers usually raise. The pricing judgement stays yours.
I'm preparing for a listing appointment at [address]. Subject property: [beds, baths, sq ft, lot, condition, updates]. Comparable sales I pulled: [paste each: address, sale date, price, beds/baths, sq ft, condition, days on market, my adjustment and why]. Active competition: [paste]. My suggested list price range: [range]. The seller's expectation: [their number and where it came from, e.g. an online estimate or a neighbour's sale]. Write: 1) a one-page plain-English explanation of how I got to the range, one short paragraph per comp; 2) a simple table: comp, sale price, adjustment, adjusted value; 3) honest answers to the 4 objections this seller is most likely to raise, given their expectation; 4) what would justify the top of the range. Use only my numbers; if a comp looks weak (old sale, very different size), say so. Don't call this an appraisal.
Set up missed-call text-back: a voicemail greeting, an instant text and two follow-ups that turn missed calls into bookings
When a small business can't pick up, many callers simply ring the next company. A missed-call text-back sends a short text right after the missed call so the conversation continues by message. This prompt writes the voicemail greeting, the instant text and follow-ups for your business type, with opt-out wording and the details to collect.
Write my missed-call text-back set. Business: [name, type, area], booking link or way to book: [link / 'reply with a time'], hours: [hours]. Write: 1) a voicemail greeting under 20 seconds, in my voice, saying we'll text them right away; 2) the instant text sent within a minute of a missed call (under 300 characters): business name, sorry we missed you, ask what they need and invite them to reply or book at [link], plus 'Reply STOP to opt out'; 3) a follow-up text 2 hours later if no reply and one the next morning, each different and shorter; 4) the 3 details to collect by text before I call back (for my type of business), asked one at a time; 5) a version for after-hours calls that sets an honest expectation for when we'll reply; 6) an emergency line, if relevant to my business, telling people what to do for urgent issues [e.g. gas smell: leave the building and call the gas utility; medical: call 911]. Keep the tone warm and local. Don't promise prices or arrival times. Note: only text people who called us, and check your local texting rules (consent and opt-out) before you switch it on.
Prep a buyer's Saturday showings in minutes: have an AI browser agent pull each listing's showing instructions from your MLS into one table
A buyer sends six homes to tour and you face six listings to open, six sets of showing instructions to copy and six listing agents to track down. An Austin agent signs in to his MLS himself, then lets Claude in Chrome open each listing and fill one table with the showing instructions and how each listing agent wants showings booked, writing "not found" instead of guessing. You still book every showing and send every message yourself.
I'm a buyer's agent. I'm signed in to my MLS in this browser. My buyer wants to see these homes on [day]: [paste MLS numbers or addresses]. Open each listing one at a time and build me a table with: address, MLS number, list price, status, showing instructions (copied word for word), how the listing agent wants showings booked (showing service, text, call or email), listing agent name and phone, lockbox or access notes, and any restrictions (notice needed, pets, tenant-occupied, time windows). Rules: never invent a fact. If a field isn't in the MLS data, write "not found". Don't book anything, don't message anyone and don't change anything in the MLS. When done, suggest a showing order that fits the time windows, and list which homes need a call or text from me first. Before using this, check that your MLS rules allow this kind of assisted browsing.
Stop chasing other agents between showings: send every listing agent all your questions at once and get the answers back as one comparison table
A coastal agent lost hours in the car waiting on voice notes and slow replies from other agencies. An AI assistant now writes to each listing agent with every question at once, sends him the key points as answers arrive, builds a table for each property, and stops to hand him anything with a catch, because he runs the deal. Start with the prompt below to write the questions and the table, and keep sending in your own hands until you trust it.
I'm a buyer's agent asking other agents about these properties for my client: [paste addresses or links]. My client's must-haves and deal-breakers: [list]. 1) Write one short, polite message per listing agent that asks every question at once, numbered so they can reply quickly: is it still available, any offers in, seller's timeline, what's included, HOA or community fees, known issues or repairs, reason for selling, showing availability this week, plus any question specific to my client's must-haves. 2) Give me a table to track replies: property, agent, date asked, each answer, still missing, and a red-flag column. 3) When I paste replies back (text, email or a voice-note transcript), fill the table, boil each reply down to three key points for me, and flag anything vague, contradictory or legally sensitive as "agent to handle" instead of answering it. 4) Draft a polite one-line follow-up for any agent who hasn't answered within [hours]. Don't send anything; I send every message myself.
More in AgentPlaybook Pro
- Personalized cold emailWrite short, specific outreach that doesn't sound like spam.
- Instant quote drafterGenerate customer-specific quotes from your pricing rules, inspired by C.H. Robinson's freight-quoting agent.
- Reach out to sign-ups no salesperson is coveringHave an agent contact registered customers who have no assigned rep, learn what they need, and pass qualified ones to sales.
- Research a prospect from their public posts and pick one thoughtful gift that gets you the meetingBusy buyers and investors ignore hundreds of cold emails, so even a good pitch never gets read. An AI agent can read what a prospect has said and done in public, find one detail that matters to them, and suggest a gift and note that feels personal, with you approving every pick and capping the spend. On Oct 21, 2026, Daymaker founder William Lindholm posted on X that the company had launched an agent that finds your prospects, reads their digital footprint and picks a gift to book the meeting, saying companies including Rippling, Salesforce and PostHog spent more than $100k with it on cakes in the previous 14 days; the post drew over 280,000 views. Daymaker's site says it reads up to 800 public sources per person, cites its sources, weighs three gift options, drops signals tied to protected traits, and buys nothing above the per-person cap you set.
- Answer resellers and new sales leads on WhatsApp around the clock, and hand a lead to a person the moment they want to orderBrands that sell through resellers or franchisees get the same questions over and over (how to join, where's the catalog, how do I pay this slip), and leads go cold waiting for a reply. An AI agent can answer these in the chat apps people already use, qualify new leads, and pass anyone ready to order straight to a human. OpenAI reported in October 2026 that Grupo Boticário runs more than 30 agents across its reseller journey: they cover 100% of app chat and WhatsApp Enterprise conversations, resolve up to 80% of operational requests on their own, and receive and qualify 90% of direct-sales leads, with the handoff rule set as 'the lead wants to place an order.' The frontline sales and support teams, not IT, own and A/B test the prompts.
- Design a phone agent that calls new leads, qualifies them and books meetings: split the call into stages, open with a hook in your brand voice, and expand only when it beats your repsMost leads that ask for a demo or quote never get a fast call back. A voice agent can phone them within minutes, find out what they need, check they fit, and book a slot with a person, but one giant prompt is too slow and too error-prone for a live conversation. Alex Lindahl shared on X (Oct 22, 2026) an interview with monday.com's VP of AI GTM, Oran Akron, describing a voice agent that calls tens of thousands of leads and books about 1,500 meetings a month: it is an orchestrator with four sub-agents (intent, qualification, commercial and scheduling), opens with a short on-brand joke because the first 5 to 10 seconds decide whether people stay on the line, keeps one memory per lead across calls and email, started with small territories and expanded only as it matched or beat human benchmarks, and is overseen by about 10 people running morning KPI checks and scoring every call and email.
- Turn conference meetings into CRM records with voice notes: after each chat, send your AI a 30-second voice memo and it logs the contact, finds their details and books dated follow-ups in HubSpotArush Jain (@Arushj01), CEO of Modus Audit, posted on X (Oct 23, 2026) that after two days at a conference, where he sent a 30-second voice note to his Grok Bot after every meeting, it had logged 39 contacts, found their LinkedIn and email, loaded them into HubSpot with 22 dated follow-ups and tracked about $6.5 million of pipeline, and he never had to open HubSpot. The same routine works with any assistant that can take voice input and connect to your CRM. Have it check before adding contacts or sending anything.
32 copy-paste agent prompts (PDF) plus 5 new prompts every week.
Get it free