# The LinkedIn library > The most useful posts by WarmUp's founders, Mike Hurley and Deisy Lewis, by topic: the workflows they run, the numbers behind them and the lessons, each with a passage of the post and a link to read it on LinkedIn. - Canonical page: https://warmup.ai/resources/linkedin-library - Mike Hurley, Founder: https://www.linkedin.com/in/warmupmike/ - Deisy Lewis, Chief Strategy Officer: https://www.linkedin.com/in/deisylewis/ ## GTM engineering Mapping a market, building the list and filling the gaps in it. ### Market mapping starts in your CRM, not in a list tool > Before building a list, read who actually took meetings: their industry, what made them raise a hand, how they described the pain. Then map the market in passes and give every company a segment and a tier. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted October 2026. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7511545967997808640/ - Impressions, as LinkedIn showed them: 805 I'm restarting something I did for a few weeks last year: a Daily go-to-market Standup. Less than five minutes a day, I'm going to share one thing I'm actually working on. A workflow, a process, a tool, a success story and something I failed or broke. People seemed to get value out of it last time, and honestly I love doing it, so we're doing it again. Day one is "market mapping". I redo this thing for WarmUp over and over, because every time I do it, it gets a little better. Here's the part most people skip: before I go to work, I often go back into our CRM. I look at who took meetings with us. What industry were they in? What made them raise their hand? And how did they describe the pain in their own words? There's what you think is your ICP, and then there's sometimes what the data in your CRM says. They are not always the same thing and for a lot of folks it's an exercise that's worth the time to do first before you start building lists. For the market mapping, I started with Clay's new audience feature (nowadays, I mostly run it from Claude Code or Codex now instead of logging in), then three or four passes to clean it up. Dead domains, industries I meant to exclude that still squeaked through, whether they even have a sales or marketing leader, do they fit our headcount, headquarters and revenue filters. Then I used a research agent to review their site, and they get a segment and a tier. Because I'm not in the UI as much now, I'm going to share a diagram with every video. Fair warning: this first one is a little hard to read, but I've linked to it below. And I went past five minutes on day one because of course I did. Video: https://lnkd.in/e9VF26Gy Diagram, with the tools at each step: https://lnkd.in/ejkxFkRE Questions or feedback, drop them in the comments or DM me. Happy to help if I can! See you tomorrow! ### Your ICP came from a meeting room. Rebuild it from closed-won data. > Export closed-won deals from the CRM with industry, size, cycle and revenue, hand the file to Claude and ask what the winners have in common. Five minutes, and the targeting stops being a guess. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted May 2026. Your ICP is wrong. Your ICP is wrong. And you probably built it in a meeting room with zero data. Here's what I see every week: Teams sit down. Brainstorm who their "ideal customer" is. Write down job titles, company sizes, industries. Then they load those filters into Apollo or Sales Nav and start blasting. 6 weeks later? 2% reply rates. Pipeline full of bad-fit deals. Everyone blames the messaging. But the messaging isn't the problem. The targeting is. Your ICP was built on assumptions. Not patterns. Here's what I do instead. Takes 5 minutes. Pull your closed-won data from HubSpot. Export it. Company name, industry, size, deal cycle, revenue. Upload it to Claude. Ask one question: "What patterns do you see in my best customers?" Read what comes back. Last time I did this for a client, Claude found three things nobody on the team expected: This is what an AI-native GTM agency does differently. We don't bring opinions to strategy sessions. We bring your data. 73% of closed-won deals were companies between 80-200 employees. Not "50-500" like the ICP doc said. SaaS + fintech dominated. The team had been targeting healthcare for months. Fastest deal cycles came from companies that had recently hired a RevOps lead. Nobody was tracking that signal. 5 minutes. No spreadsheet formulas. No pivot tables. Just patterns sitting in your own CRM that nobody looked at. The old way: founder intuition + sales team opinions + a whiteboard. The new way: your actual customer data tells you who to target. KISS principle. Upload the data. Ask the question. Let the patterns speak. Most teams are running outbound to an ICP they invented. Not one they discovered. Fix the targeting first. Then worry about the copy. P.S. If you want to try this, start with your last 50 closed-won deals. That's enough data for Claude to find the clusters. ### Build a company table from the tools a market uses > Day one of building in public: a table of telecom and IT consulting firms using HubSpot, found with the BuiltWith integration in Clay. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted March 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7304970817312821249/ - Impressions, as LinkedIn showed them: 2,285 Kicking off something new today! 🙂 I’m starting a daily standup for our Clay.com work. Part of that work will now be done out in the open. I’ll share what I’m working on, what’s working (or not), and get feedback from the community. In this first video, I describe how I built a company table targeting telecomm and IT consulting firms using HubSpot. It was my first attempt at using the BuiltWith integration in Clay... Would love to hear from other folks. Have you used the BuiltWith integration in Clay? Do you have any tips for getting better data on CRM usage? And…what should I call this daily standup series? Clay Unfiltered (raw, real work work sessions), Clay.com Daily Standup? Clay & Coffee (assuming Mike goes to sleep and wakes up when he's supposed to). Check out the video and let me know what you think! I made it using Loom. Soon this will be on YouTube (Jason Fletcher Look I did a thing). Cheers everyone. ### Fill the gaps in a list with data providers in sequence > Clay runs through several data providers one after another until the email or phone is found, which raises hit rates on a telecom list filtered to HubSpot users. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted March 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7306061581333803010/ - Impressions, as LinkedIn showed them: 678 Clay.com Daily Standup, building in public – Day 4: Waterfall Enrichment Today’s session is all about waterfall enrichment and how to systematically find missing contact details, including emails and phone numbers, using multiple data providers inside Clay. Today’s session included: - Waterfall Enrichment: A step-by-step breakdown of how I enriched a telecom company list, filtering for businesses using HubSpot, and then layering on contact details for decision-makers like VPs of Sales and Marketing. - Clay runs through multiple data providers sequentially until it finds what is needed, saving time and improving hit rates. - If you are building outbound systems, automating enrichment like this can make a significant impact and save you a ton of time. Watch today’s breakdown to see it in action. Would love to hear how others are handling data enrichment in their workflows. Let me know what's working best for you in the comments! ### From company names to domains and contacts in an afternoon > A list of several hundred target accounts with names only: ChatGPT structures it, Clay finds each domain through search, and a waterfall finds the contacts. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted July 2024. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7224451776920371201/ If you've ever had to manually build or enrich lists for your job, Clay.com is a game-changer. The time you save is great for the business and for your quality of life, too. List building and enrichment work has been a drag historically, but now it's kind of fun. Here's a super simple example: Last week, we used ChatGPT and Clay.com with a list of 650+ target accounts, in this case, agencies across the United States. However, we only had the company name but no company domain or contact information. We dropped that list into ChatGPT and asked it to organize the company name and other bits of information we had on each business into a table. Then, we added that information to a table in Clay.com and used the Google Search API to automate the finding of each company's domain. These two things (organizing the data) and, (finding company domains) used to be a slog and take a lot of time. But now it's super fast with the right tools. At this point, you can export that list and use a list-match service to match the companies in a database, then filter for the contacts you want to reach. Or, you can also use one of Clay's integrations with a single data provider, an AI agent, or a waterfall enrichment process with multiple data providers to find the right contacts at these companies. Tomorrow, I'll write a post about how the waterfall enrichment process works in Clay. If anyone wants to talk about Clay use cases and how to save a lot of time with this kind of work, give me a follow! I'll be posting about it a lot more. Until tomorrow's post, see ya and bye-bye! 👋 ## Signals Catching the people who already raised a hand: on your site, on your posts, in a job change. ### Inbound-led outbound: catch the engagement marketing already creates > Four steps that turn reactions, comments and reposts into a routed list: capture, qualify, suppress, route. Every company gets a segment and a tier before a rep sees it. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted October 2026. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7511912806208864256/ - Impressions, as LinkedIn showed them: 3,400 Happy Friday! Today's workflow is one of my favorites: inbound-led outbound. The idea: your marketing already creates engagement. YouTub, webinars, LinkedIn content, your website content. And unfortunately, a lot of that engagement is lost. This workflow catches it. Capture - One example here, use Apify through Freckle to grab everyone who reacts to your LinkedIn content, comments or reposts on posts, or on posts from a few tool companies in your industry. Qualify - Identify, enrich and score every company and person who engaged. I'm doing this in Freckle and having a lot of fun with it. We're even doing engagement scores in Supabase now, thanks Andy Toizer for sharing your incredible setup. Suppress - Remove your current clients out. Competitors out. Anyone you already have a meeting with, out, etc. Route - Every company gets a segment and a tier, then goes wherever sales can act on it: a HubSpot sequence, a task, a Slack ping, a Smartlead email, a HeyReach message, you get the idea. The takeaway: This bridges the gap between what marketing generates and what sales should be working. Sales is sooo hard. Don't miss the qualified people who already raised a hand. Want to brainstorm how this would work for your business? Feel free to DM me. I do this with one or two people most weeks, no pitch attached. Have a great weekend. See you Monday! #FreckleFriday! Freckle.io Day 2 of the GTM Daily standup series. ### Website visitors into enriched, routed leads > RB2B identifies the visitor, Clay enriches and filters for ICP fit, and qualified people go to HeyReach or an email tool with a campaign per ICP. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted March 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7306426343880044545/ - Impressions, as LinkedIn showed them: 441 End of Week 1: Daily Clay.com Standup! Happy Friday everyone! Wrapping up the first full week of doing my daily Clay.com standup videos, and honestly, I'm super proud of sticking with it. Getting started was the hardest part. I'm excited to keep this going. Today, we dove into RB2B.com + Clay and how to turn website visitors into real enriched leads. Here’s the play: - Set up a Clay table and grab a webhook. - Feed that webhook into RB2B to capture website visitors and get some initial data. - Enrich those visitors in Clay, grab emails, LinkedIn profiles, company details, and more. - Filter for ICP fit because not all visitors are worth outreach. - Push Qualified leads into HeyReach for LinkedIn or Instantly/SmartLead for email with each ICP getting its own campaign. This ensures that warm leads aren’t wasted and get personalized follow-ups. If you’re testing something similar, I’d love to hear what’s working for you. Let’s go! Have a great weekend! ## Outbound What gets a reply, on LinkedIn and by email, and what only gets ignored faster. ### AI SDRs sent more of the same message, faster > Adoption up, reply rates down. Volume was never the bottleneck: data quality, deliverability and knowing something true about the person were, and the accounts still converting did that work before any sequence went live. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted July 2026. 41% of B2B teams now run an AI SDR. Reply rates just fell to 5%. Those two numbers shouldn't move in opposite directions. But they are. We didn't fix outbound. We automated the same mediocre message and sent more of it, faster. I'll say the thing nobody wants to hear: volume was never the bottleneck. Data quality was. Deliverability was. Actually knowing something true about the person you're emailing was. AI didn't solve any of that. It just made it easier to skip. The accounts still converting right now aren't running the fanciest stack. They're the ones who did the unglamorous work first: real signals, real data, before a single sequence went live. That gap is only getting wider. AI is a lever. It is not a strategy. If your numbers are flat despite adding tools this year, the tool isn't the problem. What's actually working for your team right now? Tell me in the comments, or DM me. #GTM #B2BSales #OutboundSales #SalesLeadership #RevOps ### How we get LinkedIn connection requests accepted > A clean, verified, scored list before HeyReach touches it, then a sequence that behaves like a person: view the profile, wait, like a post, then connect. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted February 2026. We get 70% HeyReach acceptance rates. Industry average? 25%. Here's the exact workflow: 1. Build a clean list first. Before you touch HeyReach, your data needs to be perfect. Apollo for initial data, enrich with multiple providers (not just one), verify everything, score leads before you send. Garbage in, garbage out. 2. Act like a real person. This is where most people mess up. They just blast connection requests. Our HeyReach sequence: view their profile, wait, like a post (even one from 3 months ago), then send the connection request. That's what you'd naturally do. So that's what we automate. 3. No connection note. I know (The post continues on LinkedIn; only its opening is captured here.) ### Make events produce pipeline, not just leads > Pre-event workflows that book meetings before you arrive, enrichment that scores leads at the booth, and post-event sequences that reference what each attendee cared about. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted July 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7349877678465830912/ - Impressions, as LinkedIn showed them: 471 Your events generate leads. But do they generate pipeline? Most B2B teams leave 80% of event ROI on the table because manual follow-up doesn't scale. Join us at Clay Club Indy to change that. Co-hosted by, myself, the REVelry GTM team led by Joshua Kline, & Gavin Ruppert, GTM Engineer at Stellar. We're hosting a hands-on workshop where you'll learn to transform every webinar, conference, and meetup into automated revenue using Clay.com + AI. What you'll build with us: → Pre-event workflows that book more meetings before you even show up → Enrichment that scores leads while they're still at your booth → Post-event sequences that personalize at scale (no more "great meeting you!" spam) → Multi-touch cadences that actually reference what attendees cared about Real examples we'll cover: - Turn a 500-person conference list into qualified meetings immediately following the event - Automate speaker and/or partner outreach to get better response rates and create a pipeline of high powered events - Build attribution systems that prove event ROI to your CFO/CEO This isn't theory. Gavin and I will open our laptops and build these workflows with you in real-time. Where: REVelry at Nomad, 1401 North Meridian Street, Indianapolis For GTM operators, RevOps pros, event marketers, and anyone tired of manual spreadsheet hell after every event. Limited seats! Register here: https://lu.ma/ynzul9zj See you there! #EventMarketing #MarketingAutomation #RevOps #B2BEvents #GTM #Indianapolis #ClayClub #Clay ### Route qualified LinkedIn engagers into HeyReach campaigns > Pull likes and comments with Trigify, enrich and qualify in Clay, then sync to HeyReach so follow-ups run by engagement, title or industry, with email and phone layered on. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted March 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7307530504239529985/ - Impressions, as LinkedIn showed them: 674 Day 8 of my Daily Clay.com Standup: Automating LinkedIn Outreach with HeyReach Today, I walked through a cool way to automate LinkedIn outreach using HeyReach.io + Clay.com, specifically, how we’re pulling engagement data, qualifying prospects, and funneling them into outreach campaigns. Here’s the workflow we set up: ✅ Pull LinkedIn engagement data (likes/comments on my posts with Trigify.io). ✅ Enrich the data, identify job titles, company sizes, and whether they’re qualified. ✅ Sync with HeyReach, automate follow-ups based on engagement, job title, or industry. The fun part? Once the setup is done, you can auto-route them into different campaigns and even layer in email and phone follow-ups (shoutout to Ruben Trejo for making that happen!). If you're doing outbound, engagement-based workflows like this are super helpful. Would love to hear how others are automating their outreach! What tools/workflows are working for you? Letssss goooo!!! ### Eight cold email rules, three posts at a time > Subject lines of two or three words, an insight before any pitch, a colleague's tone, under five hundred characters, and follow-ups that add something new each time instead of a bump. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted 2025. 70% of cold emails get ignored in 2 seconds. You either earn attention fast or lose it forever. From our Clay Library, here are 2 tips that change everything: Subject lines: 2 to 3 words. "Quick question" beats "Q2 Strategy Proposal for Enterprise SaaS Leaders". Lead with value: not a pitch, not your company. Start with an insight or observation, something they can't ignore. Most cold emails read like brochures. That's the fastest way to get marked as spam. 2 more tips from our Clay Library: Use a natural tone: write like you'd message a colleague, not like you're begging for a call. Ideal length: under 500 characters. Think scannable. Light. Conversational. If it looks like a wall of text, it won't get read. A cold email doesn't have to convert on the first try. The real pros follow up with purpose. Here's how we think about it: Social proof: add a one-liner about your impact. Thoughtful follow-ups: don't just bump. Add something new each time, a stat, a fresh idea, a question. These are just 2 of the 8 cold email rules we include in our Clay Library. ## LinkedIn Profile visitors, likes and comments as the start of a conversation. ### Turn LinkedIn profile visitors into pipeline > A webhook pushes profile visitors into Clay, filters them against the ICP, enriches emails and titles, and triggers a personal follow-up. One run of the play led to a booked meeting and a proposal. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted March 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7305713474859343872/ - Impressions, as LinkedIn showed them: 1,590 Day 3 of Building in Public with Clay.com: Automating LinkedIn Profile Visitors into Pipeline Opportunities Today, I got some help from my teammate Ross Wimer to help me build a webhook workflow that pushes LinkedIn profile visitors from Teamfluence™ into Clay. Why is this useful? People viewing your profile might be curious about what you do, but that doesn't mean they’ll reach out. So why not meet them halfway? Here’s how we set it up: • TeamFluence captures profile visitors • A webhook sends that data into Clay • We filter visitors against our Ideal Customer Profile (ICP) • We enrich data for emails, job titles, and phone numbers • We reach out with personalized follow-ups And it works. Last week, Deisy Lewis ran this exact play, sent emails to her profile visitors, and it led to a booked meeting. That meeting turned into a proposal. Do you think this would be useful in your sales process? Let me know in the comments. ### Not every like is a lead: qualify engagement first > Trigify pulls everyone who likes or comments, Clay enriches them, and an AI prompt scores each against the ICP so only Qualified people get time. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted March 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7307222726832705538/ - Impressions, as LinkedIn showed them: 643 How to Qualify Your LinkedIn Engagement for Outbound Prospecting Using Clay.com If you're doing outbound, you know that not all engagement is created equal. Just because someone likes your post doesn't mean they’re a good fit for your business. So in today's #DailyClay standup, I walked through how I’m qualifying LinkedIn engagement using Clay.com + Trigify.io. Here’s how it works: 🔹 Step 1: Trigify monitors my LinkedIn posts and pulls in everyone who likes or comments. 🔹 Step 2: Clay.com enriches that data, company size, industry, job title, website, etc. 🔹 Step 3: I use an AI prompt to automatically qualify prospects based on ICP criteria (B2B? Senior decision-maker? Right company size?). 🔹 Step 4: The AI outputs Qualified or Not Qualified, so I’m only spending time on the best opportunities. Why does this matter? Because time is our most valuable asset! The goal isn’t just more conversations; it’s the right conversations! Hope everyone has a good start to their week! ### Track post engagement into a Clay table with Trigify > Likes and comments on LinkedIn posts land in a Clay table through a webhook, the same idea as tracking profile visits, pointed at engagement instead. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted March 2025. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7306800183311224832/ - Impressions, as LinkedIn showed them: 723 Day 6 of our Daily Clay.com Standup: Experimenting with Trigify.io + Saturday Donuts 🍩 Saturdays in my house start with donuts. It’s a tradition. And today was no different. Morning donuts, a trip to the park with the kids (until the rain hit), and now, while they nap, I’m diving into some Clay.com work. I wasn’t sure if I wanted to keep doing these standups on the weekend, but honestly, as long as it’s fun and I’m learning something new every day, I’ll probably keep rolling. Today’s Experiment: Using Trigify.io for LinkedIn Engagement Tracking Trigify.io can automatically track likes and comments on LinkedIn posts and push that data into a Clay table. The idea is similar to what we did with TeamFluence (tracking profile visits), but now we’re focusing on engagement data. Here’s the process I set up today: ✅ Created a new Clay table for tracking LinkedIn engagement ✅ Set up a webhook to capture data from Trigify ✅ Explored the fields Trigify provides, still figuring out where engagement source details are stored. ✅ Realized I probably need to watch the tutorial (we’ll blame eating too many donuts for skipping that) This type of automation looks super useful for building engagement-driven workflows. If you've used Trigify.io before, how are you leveraging it? Would love to hear what’s worked for you! Hope you’re all having a great weekend! ## RevOps The CRM work behind the pipeline: sequences, ownership, cleanup. ### A three-hour CRM cleanup in thirty minutes > A remove list with no record IDs, matched against a full company export by fuzzy matching in Claude: owner filtered, legal suffixes stripped, similarity scored, then pushed harder when the first pass came back thin. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted April 2026. I spent 30 minutes cleaning up a client's CRM. It took our last RevOps consultant 3 hours to do the same thing. Here's what changed: A client sent over a "remove" list of 118 companies that needed to be deleted from HubSpot. The problem: zero HubSpot IDs in the file. Every record showed -1. The old way would've been: manual search, match by name, hope the spelling was close, delete one by one. Instead I pulled a full export of 4,200+ companies, uploaded it to Claude, and asked it to fuzzy match the remove list against the full database. It filtered by owner, stripped legal suffixes, ran similarity scores across every record. First pass returned 8 matches. Too low. I pushed it: try harder on partial name matches, look for key words, flag anything above 70% similarity. Second pass: 19 confirmed matches. I reviewed each one manually, caught 2 false positives (wrong union, wrong foundation with a similar name), confirmed 17. Output: two clean CSVs -- contacts first, companies second -- formatted and ready for HubSpot's import-to-delete workflow. The work wasn't eliminated. The grunt work was. I still made every call on the false positives. Claude just did the part that used to eat the afternoon. That's the actual use case for AI in RevOps -- not replacing judgment, replacing the 2 hours before you get to use it. What's the last manual CRM task you're still doing by hand? ### Enrol contacts in HubSpot sequences on behalf of your reps > HubSpot assigns the sequence task to whoever enrols the contact. The fix: a Clay table as the source of truth, contacts parsed and deduped per rep, then enrolled so the task lands with the right owner. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted March 2026. HubSpot won't let you enroll contacts on behalf of your reps. (As far as I've seen, no one else has figured this out yet.) Here's the exact Clay workflow we built at WarmUp to solve it. 1. The problem with HubSpot sequences: when you enroll a contact on behalf of another rep, HubSpot assigns the task to you, not them. HubSpot's only fix: log into each rep's account individually. Not scalable. 2. The data source: we're migrating a sales team off legacy CRM onto HubSpot. Their tech team pushes the full contact database via webhook into a Clay table. That becomes the source of truth. 3. Parse by rep and dedupe: Clay routes contacts into per-rep tables automatically. No spreadsheet gymnastics. D (The post continues on LinkedIn; only its opening is captured here.) ## AI at work Where an AI assistant actually saves the week, and where it does not. ### Stop rewriting the prompt. Write the rules once, as a skill file. > Client reports, meeting recaps and task intake were being rebuilt by hand every week. The process now lives in a skill file in GitHub: what a report includes, what it never includes, how tasks get routed, where the approval gate sits. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted September 2026. I stopped writing the same claude code prompts every week. Every week I was rebuilding the same things by hand. Client reports. Meeting recaps. Task intake from transcripts into Monday. So I stopped writing the output and started writing the rules. Now the process lives in a skill file in Github. What the report includes, what it never includes, how tasks get scoped and routed, who they can be assigned to, where the approval gate sits. The first recap skill I built came from reverse-engineering a teammate's formatting. Section headings, action items grouped by owner, next meeting as a footer line. Every quirk that made it hers, written down. That is the part people skip. They want AI to do the work. The leverage is in describing the work precisely enough that it can. The by-product: I documented processes I had never actually written down. Turns out most of my operating knowledge was living in my head and nowhere else. If you are building this stuff internally, start with the deliverable you resent making the most or takes up most of your time. ### No agent needed: Claude sorted a seven-rep territory list > Pick the top manufacturing accounts from a list split across reps, proportionally, without gutting anyone's book, with the original rep kept as secondary owner. One upload and a plain-language brief instead of two hours in a spreadsheet. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted April 2026. I didn't build an AI agent last month. I just asked Claude to sort a spreadsheet. Saved 90 minutes. Here's how: The task: A client sent 1,500 companies spread across 7 reps. I needed to pick the top 200 manufacturing companies (50-500 employees) for our rep to work, while keeping the original rep as secondary owner. And I couldn't gut any single rep's book. Had to pull proportionally. The old way: Filters in Google Sheets. Manual counting. Copy-paste into separate tabs. Double-checking the math. 2 hours minimum. The Claude way: Enriched the list in Clay, uploaded to Claude, told it what I needed: → Filter to manufacturing only → Headcount 50-500 → Select 200 total, pulling proportionally based on each rep's account count → Create new column: "Primary Owner" assigned to our rep → Create new column: "Secondary Owner" with original rep name → Split into 8 files, one per rep First try it got the distribution wrong. It pulled an even number from each rep instead of proportionally. I gave it an example: "Rep A has 300 accounts. Rep B has 60. Pull more from A, fewer from B." Second try: exactly right. 30 minutes. 8 clean CSVs ready for Clay enrichment, then straight into HubSpot. Sometimes the best use of AI isn't building agents. It's just skipping the grunt work. What's a task you keep doing manually that could probably take half the time? ### Three ways Claude saves time every week, none of them an agent > Cleaning and splitting lists without a filter setup, meeting transcripts into assigned tasks with a review step, and an ICP read from a client's own customer list instead of their guess. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted April 2026. Everyone's talking about AI agents automating their entire GTM stack. I use Claude to parse CSVs faster than Clay. And honestly? That's fine. Here are the 3 ways Claude actually saves me time every week: 1. Cleaning and parsing lists You can do this in Clay. But Claude is faster, no filter setup, no re-running. Upload a CSV. Tell it what you need: → Split into 5 lists by rep territory → Group by job title or industry → Remove all .edu emails before a campaign goes out Takes 2 minutes. No workflow setup required. 2. Meeting transcripts → Monday tasks We record every planning call and client meeting. Old process: I'd spend 1-2 hours reconstructing what got assigned, to who, by when. Now: Claude reads the Notion transcript, extracts tasks with owners and due dates, checks for duplicates already in Monday, and waits for my review before pushing anything. The instructions live in a GitHub repo. Anyone on the team can use the same workflow. 10-15 minutes, start to finish. 3. ICP analysis from customer data New client comes in. They have existing customers. I enrich that list in Clay, drop it in Claude, and ask for trends. 5 minutes later: which industries they're in, which job titles dominate, where they cluster geographically. Before: we'd ask clients "who do you think your best customers are?" They guessed. Now: we show them the data. --- If you want the prompts I use for #2 and #3, comment "CLAUDE" below. ### Meeting transcripts into assigned Monday tasks, one command a day > Recordings land in Notion during the call. Once a day, one command has Claude read the week's transcripts, separate client meetings from internal calls, extract the action items and create the tasks on the right board. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted February 2026. I used to spend 30+ minutes after every client meeting creating tasks manually. Open Notion. Find the action items. Open the right monday.com board. Create tasks. Assign them. Set due dates. Check for duplicates. Multiply that by 4-6 meetings a week. So I built an automated workflow. Claude + Notion + Monday.com. One command per day. Here's what it does: Meeting transcripts go straight into Notion. Press a button during the call and the recording lands in a dedicated page. Once a day, I paste one command into Claude: "Process this week's meeting recordings." Claude reads every transcript from that week. Identifies which ones are client meetings vs. internal calls. Then extracts action items, (The post continues on LinkedIn; only its opening is captured here.) ## How we work The firm, in the founders' own words. ### Six months before admitting failure: how WarmUp started > The bet, the pivot, and the lessons: find people smarter than you and pay them, expect pain, endure, and treat everyone like humans. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted February 2026. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7424440723724238848/ - Impressions, as LinkedIn showed them: 3,227 I gave myself 6 months before admitting failure. That was nearly 7 years ago. In 2019, the consulting company I worked for, Fretless, was being acquired. I had a choice: take the safe job, or bet on WarmUp.ai full-time. I gave myself 6 months. If it couldn't pay the bills by then, I'd go get a real job. No shame in trying, right? Honestly, that was the scariest 6 months of my life. It worked. Barely, but it worked. Here's what years of building has taught me: I spent years in sales solving "how do I close this deal" sort of puzzles. Building a company taught me to enjoy a different level. How do you position a business to attract the right customers? How do you build systems that create opportunities consistently? The deal puzzles are still fun. But the go-to-market engineering (GTME) ones are where you create real leverage and they're super interesting. Find people smarter than you and pay them. Dan Rosenthal has taught me more about GTME work than any course or book. I literally asked him: "How much do I have to pay you to be my friend every month?" He kindly laughed. I was serious though. The opportunity cost of not knowing what you should know is a compounding drag. Prepare to be in pain but trust it'll get better. My former boss Brian Rowe used to say: every day you don't quit, you're one day closer to not dying. (I love Brian mostly because he reminds me of a pirate captain ) - What he said might sound dark. But it's true. The ability to endure might be the most important thing. In late 2020 we actually had to pivot the entire business model. SaaS to agency in 6 months. It was hell. A complete rebuild. Then we had to learn Clay.com years later and learn how to sell it, another huge mountain to climb that Deisy Lewis led. The harsh truth is most people quit before they make it. Some get really close and still quit. The market for what we do now is the strongest I've ever seen. GTM engineering is exploding. But the best part of surviving this long? It's actually pretty fun right now. Interesting problems. Great clients. A team I genuinely like working with. That combination is rare. Everyone knows we don't get all three of those things most of the time. Treat everyone like humans. I've tried to run WarmUp with one central mantra: treat everyone like humans. Miles Sterrett and Davey Strus embodied this during my years at Fretless and I carried it forward. It sounds simple until you realize how many decisions it actually impacts. Employees. Clients. Partners. Everything. I haven't been perfect. But that mantra has guided my decisions. And I've ended up surrounded by people I actually want to spend time with. That's the thing about building. If you do it long enough and don't compromise on the human part, work stops feeling completely like work. If you're building something right now and wondering whether to keep going: Keep going. And if you want to chat about your journey, reach out. Do what I did with Dan. ### From one hour of cold calling a day to Chief Strategy Officer > Interpreter, medical coder, photographer, business owner, then a parent who wanted to work one hour a day. The first client stayed three years. What to look for in a company and a business partner. - By Deisy Lewis, chief strategy officer of WarmUp (https://www.linkedin.com/in/deisylewis/). Posted January 2026. I started my career as a legal interpreter. Then medical interpreter. Medical assistant. Medical coder. I ran a wedding photography business (still run it actually). A furniture restoration business. My ex-husband's construction company. Then I had my daughter. And I wanted to work one hour a day. That's when I found Mike Hurley & WarmUp. One hour of cold calling. That was the job. My first client? They stayed with us for 3 years. One hour became two hours. Two hours became manager. Manager became Chief Strategy Officer. Here's what I've learned about finding the right company and business partner: 1. They see potential, not just your resume. I had zero go-to-market experience. I had a dozen (The post continues on LinkedIn; only its opening is captured here.) ## Community The Indianapolis GTM meetups and what they are for. ### A meetup format for people who build GTM systems > Three builders show a system on screen with the number it moved and what broke, twenty minutes on someone's stuck problem, no decks, any tool. - By Mike Hurley, founder of WarmUp (https://www.linkedin.com/in/warmupmike/). Posted September 2026. - On LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7511167700321165313/ - Impressions, as LinkedIn showed them: 834 Indy GTM professionals! I'm going to get back into hosting a monthly meetup for the people who are build (or want to learn how to build) GTM systems. Next event will be on October 28, called the Indy GTM Meetup. If you can't make this one sign up for the calendar or DM me and I'll make sure you get an invite to the next one. Here's the format I'm thinking about running: 1) Three people show a system they run, on screen, with the number it moved and what broke along the way. You show us what you built. Show us what worked, what it cost, etc. 2) Then twenty minutes where someone brings a problem they're stuck on and the room works it. 3) No decks. Any tool. Clay, Freckle, HubSpot, Salesforce, Claude Code, Codex, a spreadsheet, whatever you run. I'll go in October, I've got the three workflows that we run at WarmUp every morning: a hiring-signal feed, a social listening play, and the ICP map they both read. I'll share everything. The event will be on Wednesday, October 28th, 5:30 to 7:30. Pizza and drinks on us since everyone will be hungry right after work. I'm going to open up 40 seats. If you run GTM systems in Indy, or you want to, please come join us. Or if you've got a build worth showing, let me know, I need people who want to present in November and December. https://luma.com/m3sfry00