All plays

• 01

  For sellersProspecting & researchClaude + BlitzAPI (job-postings data)

  Turn a target account’s job postings into your reason to reach out

  self-reported

  10–12 meetings a month for a Series A cybersecurity client, from Claude reading the job postings of a 5,000+ account list. He says the same system books 30+ meetings a month for a software development firm.

  He pulls every job posting since January 1, 2026 for a 5,000+ account list, and a Claude skill decodes each one: the tools named in the requirements, the pay, who the role reports to and what it’s meant to fix. Accounts get ranked on fit, movement and budget, and each comes with a draft first message that opens with what the hiring says, not with a pitch.

  By Jainendra Ojha, outbound and GTM builder, posting on LinkedIn · Source ↗

• 02

  For sellersFor sales leadersCold emailClaude Code + Instantly or Smartlead API

  Mine your old cold email campaigns for what actually booked meetings

  self-reported

  On one client account with 300,000+ emails sent, a single Claude pass over every campaign showed that pain-point openers and certain calls to action were behind most of the booked meetings. Nobody on the account knew until then.

  Instead of guessing at new copy, he has Claude pull every campaign from the sending tool (the copy plus sent, replies and meetings booked), including the flops. Claude maps copy to outcomes with four questions, must cite the campaigns behind every claim, and turns what worked into 3 to 7 named frameworks. The next campaign tests one change at a time.

  By Kenny Damian, Head of GTM at Frontal AI · Source ↗

• 03

  For sellersFor sales leadersCall prepClaude Cowork (scheduled skills)

  Put your call prep on a schedule so the brief is waiting before every meeting

  as reported by the vendor

  About 90 minutes a day saved from small scheduled tasks, including a pre-call brief for every customer meeting, plus about 3 hours a week on the Friday forecast, for a leader running a 4,000-account book.

  Each morning a scheduled skill reads his calendar and builds a brief before every customer meeting, pulling spend data and pipeline status from Salesforce, so it’s waiting when he opens his laptop. He says the schedule mattered more than the skill: once prep runs on its own, he stops forgetting it.

  By Travis Bryant, Head of US Mid-Market GTM at Anthropic · Source ↗

• 04

  For sellersFor sales leadersDiscoveryClaude (free MEDDIC skill from Stage 2 Capital)

  Score every discovery call against MEDDIC, for yourself or your whole team

  as reported

  From one call transcript: a scored MEDDIC scorecard, manager-style commentary, ranked next steps and a draft mutual action plan email in under 2 minutes. On one portfolio company’s call it caught that the rep heard who the real decision-maker was and never asked how to reach him.

  Reps tend to fill in MEDDIC optimistically, and managers don’t have time to listen to every call. Stage 2 drops the transcript into Claude with a debrief skill that scores each element as Strong, Partial or Missing with evidence from the call, names the moment the seller should have pushed, and drafts the follow-up. Their short prompt: “Analyze this discovery call against MEDDIC and give me my next steps.”

  By Mandy Cole, writing for Stage 2 Capital’s GTM newsletter · Source ↗

• 05

  For sellersDemoClaude Code

  Build each prospect’s demo from their discovery notes, and test it before the call

  self-reported

  Building a prospect-specific demo went from 3 to 4 days to 3 to 4 hours. Across his whole AI setup, his two-person team managed $31.8M in pipeline over seven months, tracked 199 demo meetings and closed $1.95M.

  Every healthcare prospect had different systems, payers and pain points, so generic demos fell flat. He built a demo skill that takes the prospect’s context from discovery (specialty, systems, deal stage, the workflows they care about) and produces a tailored demo package plus a test checklist. If the demo can’t pass its own tests, he knows before the call, not in front of the CIO.

  By Dan Mathieson, Director of Sales Engineering at Smarter Technologies (healthcare software) · Source ↗

• 06

  For sellersFollow-upClaap (call recorder) + Claude artifacts

  Turn the discovery recording into a same-day proposal in the buyer’s own words

  self-reported

  Proposal sent 2 hours after the discovery call instead of her usual 2 to 3 days. The prospect, who was comparing three agencies, replied that afternoon (“This is exactly what we need”) and signed by end of day.

  She pulled the call recording into Claude and asked for every objection, pain point and quick win, plus the exact words the buyer used. Then she built an interactive proposal page mapping their challenges to her plan, with proof, pricing and timeline, and a comparison built from the objections they raised. She sent both with a short walkthrough video while the call was still fresh.

  By Romana Kuts, at the B2B SaaS marketing agency SaaStorm · Source ↗

• 07

  For sellersNegotiation & closingClaude

  Test a late-stage discount demand before you answer with a number

  self-reported

  A $490K deal closed at full price, with zero discount, 24 hours after he countered procurement’s 20%-off demand with a 3-year price lock and net-60 payment terms.

  Procurement showed up at the end asking for 20% off, and the champion said they were “firm.” Before replying, he gave Claude the full deal history, the procurement contact’s public LinkedIn and the champion’s exact words, and asked whether this was a real budget limit or a tactic. The read: low internal pressure and no budget ceiling ever mentioned. So he offered cash-flow and budget certainty instead of price.

  By Idan Baum, posting on LinkedIn · Source ↗

• 08

  For sales leadersCRM hygieneClaude Code + HubSpot MCP connection

  Clean your CRM in an afternoon, but make AI check for open deals before anything is deleted

  self-reported

  450 companies cut to 73 in one afternoon (about 3 hours), with every remaining record given an owner, and two companies with open deals caught just before they would have been deleted.

  His HubSpot had 450 companies, lots of duplicates and no owner on half of them. He connected Claude Code to HubSpot and worked through it in order: summarize the mess, merge duplicates, check deal links before flagging anything, fill gaps and assign owners, then delete the junk. Claude stopped him on two “dead” companies that still had open deals.

  By Matteo Lombardi, founder of the consultancy Stratega · Source ↗

• 09

  For sellersLinkedIn & social sellingClaude Code

  Turn your sales calls and voice notes into a week of LinkedIn posts

  self-reported

  12 meetings booked from LinkedIn in 7 days, from an account with under 1,100 followers.

  Thirty minutes a week of raw input (voice clips on walks, sales call recordings, screenshots of posts that made her stop scrolling) goes into Claude Code, which already holds her positioning, buyer, objections and best past posts. It comes back as 10 ideas; she picks 4 or 5, writes for four different buyer stages, uses a useful giveaway to turn comments into conversations, and spends 30 minutes commenting on other people’s posts before each one goes out.

  By Kanchan Bhatta, distribution and GTM, posting on X · Source ↗

• 10

  For sales leadersCold emailWhat not to doAI SDR / autonomous outbound tool

  Don’t let an AI SDR send 1,000 unchecked emails a day from your main domain

  self-reported lesson

  After running AI outbound at about 1,000 emails a day, his company’s own invoices, password resets and deal confirmations started getting rejected by Gmail. Live emails were also citing things that never happened, like a prospect’s “recent migration to Salesforce.”

  The first month looked great: lots of replies. Then a customer asked why they’d never been billed. The outbound ran from the same domain as their invoices, and once spam complaints climbed, Gmail started turning that domain away. When he finally read the live sent emails (not the approved samples), he found confident, perfectly written claims that weren’t true.

  By Ashutosh Singhal, founder of Veriprajna, writing on Substack · Source ↗

• 11

  For sellersFor sales leadersProspecting & researchClaude Code + Google Maps data (RapidAPI) + Exa + MillionVerifier

  Build your whole market list wide first, then let AI check every website

  as reported

  25,763 companies checked and 14,710 kept as the market. 198 of 200 in the final spot check were correct (99%). 1,591 shops showed a hiring signal and 743 made the first campaign list. Tool cost was $230.

  Database filters miss real customers and let in look-alikes. Jan’s team searched wide (every label their best customers use, ZIP code by ZIP code on Google Maps, plus manufacturer partner pages), removed duplicates, then had AI read every company’s website and answer one yes/no question with a quoted line of proof. Only after that did they look for signals and contacts.

  By Jan Berning, Head of Growth and GTM engineer, writing an X article · Source ↗

• 12

  For sellersFor sales leadersCold emailClaude Code (drafts), human check before send

  Swap AI-written “personal” first lines for one real detail from their business

  as reported

  Their AI-written personalized first lines got a 0.27% reply rate. A campaign built on one real detail from the prospect’s business (their project and location) got 14.5%.

  Everyone has the same AI now, so prospects get dozens of cold emails a day with the same fake-personal opener. His team still lets AI draft every email, but a person checks every line, and the message is built around one true, specific fact about the prospect’s business instead of a generated compliment.

  By Christian (@coldemailchris), founder and CEO of Pipeline, posting on X · Source ↗

• 13

  For sellersLinkedIn & social sellingAny chat AI (ChatGPT, Claude, Grok)

  Earn the LinkedIn reply with a connection note built on one verified fact

  as reported

  In September his LinkedIn messages got 80x the response rate of his emails.

  Give AI only verified facts about the buyer and what you can truly offer. It drafts a short connection note and a follow-up; you review them and send them yourself.

  By Carson V. Heady, sales leader and author, posting on X · Source ↗

• 14

  For sellersFor sales leadersCold emailClaude

  Test AI first lines against a plain template before you pay for them

  as reported

  Over 4 months: Claude-written first lines got 6.8% replies (3.1% positive); templated problem-frame openers got 7.2% (3.4% positive). AI personalization cost 2 to 3 hours per 1,000 leads vs about 20 minutes per campaign for the template.

  Before spending hours on AI-personalized openers, split-test them against one sharp problem-frame template. In his data the template won; his takeaway is better lists, not better openers.

  By Jason Park, founder of Frostmailer (cold email agency), posting on X · Source ↗

• 15

  For sellersDiscoveryClaude + Gong export

  Mine your own last 300 calls for what you do when you win

  as reported

  From 297 of her own calls: on won calls, customers told stories 70% longer (117 seconds vs 69), asked 25% more questions, and her talk ratio was 9 points lower. Her win rate went from 38.7% in 2024 to 45.8% in 2025.

  Nevena exported her call stats with outcomes, added a few full transcripts, and had Claude compare won vs lost calls and the old her vs the new her. The numbers show the trend; the transcripts show what she actually does differently.

  By Nevena Tomovic, Lead Account Executive at Tenderly, posting on LinkedIn · Source ↗

• 16

  For sellersFor sales leadersCoachingClaude / Claude Code, Notion, AI role-play

  Build a golden call script from won and lost transcripts

  as reported

  AE win rate rose 8 points, from 29% to 37%, in under a year. 212 BDR transcripts were analyzed in about 3.5 minutes.

  Pull won and lost call transcripts, have Claude find what winners do differently, turn it into one script, and make reps pass an AI role-play on it before they get leads.

  By Marchelle Mooney, VP of Sales at Mangomint, interviewed on the GTM AI Podcast · Source ↗

• 17

  For sales leadersNegotiation & closingDust agents, Momentum (automatic CRM updates)

  Have AI draft the CFO email so the no comes on day 3, not day 73

  as reported

  ARR per rep 3x, average deal size up 40%, mid-market win rate doubled, selling time from about 25% to nearly 50%, and reps carrying about 40 deals instead of 20.

  First ask each rep to walk you through yesterday and automate the non-selling work. Then have an AI agent draft short emails to the CFO or CEO on every live deal; reps check the facts and leaders send, so you hear no on day 3, not day 73.

  By Ryan Milligan, CRO of QuotaPath, interviewed on the GTM Science podcast · Source ↗

• 18

  For sellersFor sales leadersCoachingClaude Code

  Build a cold call coach that reviews every call that didn’t book

  as reported

  Connect-to-conversation rate went from under 30% to over 60% (82% one week); 33 phone meetings booked over about 5 months.

  After every connect that didn’t book, ask AI three questions about the transcript. Volume plus direction moved his numbers; he later scaled it into a team coaching agent.

  By Won You, BDR, then Growth Lead at Great Question, posting on LinkedIn · Source ↗

• 19

  For sales leadersCoachingAI call scoring (Replays)

  Score calls, find the skills that win, and coach one at a time

  as reported

  Win rate went from 38% to 52% in about one quarter (the host says 53%) after training and daily scoring of end-of-call skills.

  Score calls with AI against many skills and find which ones go with wins. For Owner.com, 10 of 12 were in the last 5 minutes of the call: pricing, value stack, next steps, asking for the business. Then train and coach one skill at a time.

  By Kyle Norton, CRO of Owner.com, interviewed on the GTM Science podcast · Source ↗

No plays match that filter yet.