The Goats of Growth

Evan Leek says good AI outreach is built like an Oreo: a human up front, AI in the middle, and a human pass at the end. That last layer is where his team catches the "dog whistles" that flag a message as AI, and he says buyers are getting a lot more perceptive to them.Evan is Senior Director of Integrated Marketing at Aqurio, an agentic AI company for healthcare and other industries where getting it right isn't an option. Aqurio launched this summer as an AI-first company. Evan runs growth marketing and manages the sales development team, and the company tracks AI usage internally on a leaderboard.His central argument: the tech stack hasn't changed, the method has. His team still runs on the same cadence tools and CRM, with Claude layered on top. AI does the research that lets SDRs localize outreach to each subvertical and persona. It doesn't write the cadences. Every sequence gets a human review before it goes live, and he deliberately stops short of automating sends, because in marketing everything you do is public.The proof: one of his reps is running a 53% open rate on a cadence. He says it came from the work, not the tool. She studied the cohort, thought hard about her messaging, and stayed on top of every touch point. When the hook lands, he says, meetings happen within days instead of weeks.WHAT WE COVERThe Oreo cookie approach: human, AI, humanSpotting the "dog whistles" that make outreach read as AILocalized and prescriptive: the two words he holds outreach toInstitutional knowledge you can't train a model onWhy growth marketers belong on pipeline callsA 53% open rate, and how a clear next step shortens sales cyclesTracking AI usage with an internal leaderboardA team culture of sharing the agents people buildWhy he keeps a human wall before anything goes liveThe hiring assignment: 24 hours, use AI, and he grades the processAI FOMO, and why marketing is a cruise ship, not a speed boatRapid Fire Five, Scott Hanson's bonus question, and Evan's question for the next guestCHAPTERS0:00 Intro and Common Room1:09 Aqurio and Evan's role1:50 Running an AI-first go-to-market team3:26 Using AI to localize SDR outreach5:00 When the agent sounds like an agent6:00 The Oreo cookie approach7:40 Localized and prescriptive10:05 Institutional knowledge a model can't learn11:44 Staying embedded with sales13:31 Open rates in a crowded inbox16:00 The 53% open rate16:21 Shorter sales cycles17:44 A word from Common Room18:14 Managing a team with AI20:46 What an amazing use of AI looks like24:13 A culture of sharing25:17 The tech stack hasn't changed26:16 Keeping a human wall before anything goes out27:55 The skills that win in hiring now29:01 The 24-hour AI assignment31:38 What's next: getting control of the inbox33:45 AI FOMO and the cruise ship36:51 Rapid Fire Five38:59 Bonus question: if you knew you could not fail39:56 Evan's question for the next guest41:55 Where to find EvanConnect with Evan Leek on LinkedIn, or learn more about Aqurio at aqurio.com.This episode is sponsored by Common Room (https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.New conversations with go-to-market leaders every week. Subscribe to The GOATS of Growth on Apple, Spotify, or YouTube.

Scott Hanson says chatting with ChatGPT in a browser isn't AI fluency. To him, AI fluency means using AI to build a tool that another person uses to support themselves. He says it's far more attainable than most people realize.Scott is the founder of Tenfold Labs, which builds AI systems for Fortune 500 teams. Before that he co-founded Pop Bookings, the "Uber of event staffing," as COO and a non-technical founder. The app has roughly 400,000 active staff. He started building apps with AI in 2022 and was one of the first power users of Rork.His central argument: keep a human in the loop. He skipped black-box agent setups like OpenClaw and works in Claude Code and the CLI, watching what the agent does until he trusts it. He uses agents for narrow jobs inside plain, repeatable software, and doesn't let them own whole workflows end to end.The story: one conversation about AI at his Austin gym led to a contract with a $50 billion Berkshire Hathaway subsidiary in distribution. Its reports took weeks to compile and reached the C-suite one to three months late. Scott built real-time "transparency layers" across hundreds of initiatives. He says Deloitte would likely have charged around $1 million and taken eight months to a year. He delivered in two to three months, as a solo developer, for a fraction of that.WHAT WE COVERWhy he skipped OpenClaw and went human-in-the-loop in the CLIWhat a CLI is, explained for non-developersWatching the agent work: the mechanic metaphorA personal AI harness for email, calendar and financesReal-time reporting for a $50B companyHow AI exposes the dark areas of a company, and what that means for middle managementHis definition of AI fluency, and starting with the work you hateTurning workflows into simple software instead of black-box agentsMatt Pocock's "grill me" skill and a Fathom-to-CRM workflowRapid Fire Five, Allyson Barr's Mars question, and Scott's question for the next guestCHAPTERS0:00 Intro and Common Room1:12 From Pop Bookings to AI3:43 Early ChatGPT and building his own apps5:50 Finding Rork7:42 Why he skipped OpenClaw8:38 What a CLI is9:59 Human in the loop vs. the black box12:20 Build trust by watching the agent work16:12 Building his own AI harness17:38 The conversation that became Tenfold Labs20:46 The $50B client and real-time transparency layers24:24 Hundreds of initiatives, and what it's worth26:45 AI, headcount and middle management28:49 What AI fluency actually means29:44 Start with the work you hate31:02 Breaking down a workflow36:37 When a workflow should become software37:50 Let Claude interview you39:15 Go basic: do one thing42:38 Rapid Fire Five46:35 Bonus question: would you go to Mars?47:37 Scott's question for the next guest49:02 Where to find ScottConnect with Scott Hanson on LinkedIn, or learn more about Tenfold Labs at tenfoldlabs.io.This episode is sponsored by Common Room (https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.New conversations with go-to-market leaders every week. Subscribe to The GOATS of Growth on Apple, Spotify, or YouTube.

Steve Wooledge didn't build a bigger marketing team. He built a shared AI operating system, and Collate no longer has a social media hire or a customer case study writer. The contractors who covered that work came in house.Steve is CMO at Collate, the semantic intelligence platform built on open metadata. He has spent 20 years in data and analytics marketing at Alation, Alteryx, Hazelcast and Teradata. Collate is the commercial backer of the Open Metadata Project, and he explains why a context layer matters when AI agents query company data.His central argument: AI is only as trustworthy as the context it can see. He says Claude Opus pointed at a data warehouse gets roughly 60% of answers right, and Collate's benchmarks with context reach 94 to 95%. The same idea runs his own marketing org. The team keeps its skills, messaging and memory in one GitHub repo and updates it through pull requests with peer review, about 50 in the last six months.The concrete number: Steve puts AI at roughly 20% of his marketing budget saved, and thinks 30 to 33% is reachable. He's clear that's a gut estimate, not a measured one. We also get into where he reinvests it (organic over paid search and social), how his SDRs use Common Room signals, and the 7:30 a.m. email his AI second brain sends him from every call.WHAT WE COVERWhy AI answers on structured data run about 60% accurate without contextHow Collate's context layer gets benchmarks to 94-95%A team GTM operating system in GitHub: skills, messaging, pull requestsCursor for non-coders, and why he keeps his "brain" localThe roles Collate didn't hire after bringing content in houseThe 20% budget saving, and why he pushes it toward 30%How he evaluates AI tools: referrals, not demosFathom call summaries feeding a daily action-item emailRapid Fire 5, Jessica Storry's bonus question, and Steve's mulligan question for the next guestCHAPTERS0:00 Intro and Common Room1:26 How Collate uses Common Room signals4:16 What Collate does: a context layer for AI7:11 From AI noob to the "AI CMO"10:22 Cursor, explained for non-coders13:41 Whisperflow and Friday team AI sessions16:14 Claude Code from your phone17:41 What AI did to marketing velocity and hiring20:03 Quantifying the savings: 20% to 30%24:19 Where the savings go: organic over paid26:05 Sales cycles and deal velocity26:58 How Steve evaluates AI tools29:13 Why human content matters more now30:09 The Fathom second brain and 7:30 a.m. email32:50 The shared GTM OS: pull requests and peer review35:32 Rapid Fire 542:36 Bonus question: starting over, not in tech44:08 Steve's question for the next guest45:47 Where to find SteveConnect with Steve Wooledge on LinkedIn, or learn more about Collate at getcollate.io.This episode is sponsored by Common Room (https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.New conversations with go-to-market leaders every week. Subscribe to The GOATS of Growth on Apple, Spotify, or YouTube.

Dustin Schwarz's team replaced PayReady's human concierge outreach with a fully AI solution. It underperformed: skilled people reach fewer residents, but they resolve far more of the conversations they do reach. So PayReady built a hybrid process where automation handles routing and simple conversations, and specialists take escalations and sensitive discussions. Schwarz is chief customer officer at PayReady, the multifamily SaaS platform behind AI-driven rent collection, resident and payment portals, and CRM tools serving more than three and a half million units. He has spent 15-plus years in multifamily and prop tech, including time at RealPage. His argument: AI is the worst it will ever be today, but "good enough" is a business judgment, and many companies will accept 85% of a human's effectiveness because the incentives reward cutting cost over serving customers. PayReady is betting on the opposite, a customer-centric 100%. That bet is backed by a chief risk officer, a vendor network that shares regulatory updates, and a view that SaaS built on a feature is dead while SaaS built on partnerships and judgment is not. One concrete result: after buying Buzz.ai, an AI platform trained on property management behavior, PayReady saw variable margin improve roughly two and a half to three times, and one support agent can now serve about three times as many customers. WHAT WE COVER- Why a fully automated concierge replacement underperformed human specialists- Where humans stay in the loop, and whether that share shrinks over two years- State and local contact rules, junk-fee scrutiny, and legislative risk- Why "85% good enough" wins in some companies and what that means for customers- Build versus buy: acquiring Buzz.ai, and moving support to Intercom- Hiring for adaptability, creativity and language skill over domain experience- Where AI still fails: people, judgment and high-stakes analytics CHAPTERS00:00 Why this episode is out of order00:57 Intro02:21 What PayReady does04:10 Buying Buzz.ai06:41 Why the all-AI rollout fell short09:24 The hybrid process10:40 Will human-in-the-loop shrink?12:15 Legislative and regulatory risk14:30 Staying compliant17:03 Why feature-based SaaS is dead18:21 Why 85% is "good enough" for some companies21:06 Business impact: margin and support22:08 Build versus buy27:05 Hiring for adaptability over domain experience34:58 Where AI still fails39:19 Rapid Fire Five Connect with Dustin Schwarz on LinkedIn, or learn more about PayReady at payready.com. This episode is sponsored by Common Room (https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI. Subscribe to The Goats of Growth on YouTube, Apple Podcasts and Spotify, and leave a rating or review so more go-to-market leaders can find the show.

Allyson Barr spent the first six months of this year watching her content team hand back sixty to seventy percent of every AI draft for rewrites. That number keeps dropping now — not because the AI got smarter on its own, but because her team built it an editorial memory: voice dossiers on every subject-matter expert, a running log of what not to say, and a habit of feeding every human correction back into the system. Allyson is the Chief Marketing Officer at Active State, which has been securing open source software since 1997 and now sits squarely in front of the problem it was built for: AI coding assistants pulling unvetted packages into production. She joined to run a full commercial transformation, rebranding and relaunching the company in five months, after three years at AWS running sales planning across a hundred-billion-dollar business. She's also, as of this summer, co-founder of a second company, Culture Flow. Her argument: AI didn't just speed up Active State's content output, it changed where marketing spends its time. Six months ago the company loosened its AI guardrails company-wide while tightening standardization around Claude as the primary assistant, and content volume jumped from roughly six pieces a month to about a hundred. The deeper shift, she says, is that the team's real job stopped being production and became curation.The clearest example she gives is a cold email. A vendor pitching Active State built their entire outreach around the titles from a Taylor Swift album she'd posted about — a few minutes of real attention that landed harder than any AI-personalized blast, and the one she now points to when she talks about the line between a trust enhancer and a trust eroder in AI-assisted outreach.WHAT WE COVERTraining AI on five separate voice profiles for Active State's subject-matter experts The February shift that loosened AI guardrails while tightening standardization Measuring AI's impact: content volume, SDR research time, and AE proof-of-value cycles Build vs. buy in an era of vibe-coded internal tools What she looks for in candidates now that "using AI" is table stakes Culture Flow, her side venture building a workplace-trust tool CHAPTERS0:00 Intro & Common Room1:27 Active State: securing open source3:26 Building AI trust in go-to-market4:36 Training AI on the company's voice8:44 The February AI philosophy shift10:10 Measuring AI's commercial impact12:15 Scaling content across channels13:23 Automating workflows with Google Drive and Claude16:47 Prioritizing the AI initiative backlog21:41 Automating the podcast workflow: a real-world example24:09 Cutting through tool noise with personalized outreach27:15 Trust erosion vs. trust building in sales28:37 AEO and build vs. buy33:13 New skills, hiring, and the AI generation gap37:50 Culture Flow and the Rapid Fire FiveConnect with Allyson Barr on LinkedIn, or learn more about Active State at activestate.com.This episode is sponsored by Common Room (https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.New conversations with go-to-market leaders every week. Subscribe to The GOATS of Growth on Apple, Spotify, or YouTube.

When Aware launched a new product this year, Brian Krause didn't pull SDRs off existing accounts or hire three more. He put top-of-funnel demand gen on an AI agent from day one, 100%, on purpose. New product, new ICP, no legacy SDR team to muddy the results: "It's the perfect scenario to start at zero and just build an agent for it." What he wants out of it is clean data on which of the platform's value propositions actually land with buyers.Brian is CRO at Aware, the first US-based biometrics company, 35+ years in, where he runs sales, account management, customer success and RevOps. Before that he led revenue teams at identity verification and computer vision companies, and at Pivot3, later acquired by Quantum.The agent is "a campaign manager on steroids," running email and LinkedIn, but nothing goes out without a rep's eyes on it. Two reasons. The handoff to a human AE can't feel jarring. And on a technical product, the agent can't say yes to something that's only true if the customer spends more money. On hiring, he's blunt: "If I have to teach you what a CRM is, I'm not okay with that." Reps should show up AI-enabled. Aware's new head of federal runs on tooling about 99% of the day, and she works at a pace well above everyone else.He also has the most honest take on CRM hygiene you'll hear: every company has the problem, and the fix isn't enforcing it harder. It's taking the ask away from reps entirely.WHAT WE COVER- Why a brand-new product launch is the right place to test an AI SDR- Isolating variables: what the agent's data tells product about which value prop wins- The two guardrails: brand-consistent handoffs, and no overpromising- Build vs. buy: buy customer-facing, build internal productivity- RevOps as the owner of the agent roadmap- Define what good looks like before you automate it- Why hiring is plateauing, and what BDRs and AEs should do about it- Rapid Fire: three teenage daughters, and ringing the bell on Wall StreetCHAPTERS0:00 Intro and sponsor: Common Room1:49 Brian's role at Aware3:52 How trust gets built into biometrics5:30 Sales is still a people business7:43 Where AI sits in the funnel today9:24 Going 100% AI SDR on a new product15:57 Who built the agent, and how it was trained18:03 A campaign manager on steroids19:47 Sponsor: Common Room20:14 The two guardrails23:50 Build vs. buy26:34 RevOps owns the agent roadmap, and the CRM hygiene fix28:55 Define good before you automate30:49 Hiring plateaus, and showing up AI-enabled33:48 "If I have to teach you what a CRM is…"37:07 Rapid Fire Five40:31 Ringing the bell on Wall Street41:37 OutroConnect with Brian Krause on LinkedIn, or learn more about Aware at aware.com.This episode is sponsored by Common Room (https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.To hear more conversations like this one, subscribe to The GOATS of Growth on Spotify, Apple Podcasts, or YouTube.

Jessica Storry co-founded Stockpress in 2019 out of the web design agency she was running, drowning in files across Dropbox, Box, Google Drive and hard drives clients mailed them. They took a contract building on an open-source DAM, decided it was overcomplicated, pitched the customer on something simpler, got told no, and built it anyway -- on her promise that she'd sell it. She did. Stage 2 Capital invested, they closed the agency in 2020, and Stockpress now occupies the gap nobody was serving: a $9.99 Dropbox account on one side, a $2,500-a-month enterprise DAM on the other, nothing in between.The same instinct drove the decision that actually grew the business. They started out cloning the enterprise DAMs and charging less, and were still doing every onboarding by hand. So they re-architected around self-serve, lowered the price -- the only time it's ever gone down -- and replaced a product full of in-app pop-ups with a five-step onboarding checklist. "Maybe we don't give customers enough credit sometimes. They can figure it out if they if it's presented in the right way." She also tells the story of the Friday night she spent topping up tokens in Claude Design and ended with nothing to show: "I could have just designed that page myself in an hour and it would have been fine."WHAT WE COVER- Arguing against an AI support agent, and being wrong: 72% of tickets closed by the bot- What AI is allowed to write, and what has to be accurate enough that it isn't- Wireframes straight to the dev team, and why nothing AI-made ships as final design- Going self-serve: lowering the price and cutting the hand-holding- The five-step checklist that beat in-app pop-ups- 85% of customers leaving every AI feature switched on- Buyers who arrive from ChatGPT having never seen the website- Where AI burned money for an eight-person team and produced nothingCHAPTERS0:00 Intro and sponsor: Common Room1:16 What Stockpress is, and the founding story4:23 Cutting the hand-holding, and lowering the price7:59 The five-step checklist that replaced the pop-ups10:28 AI inside the product, and the 85% who turn it all on13:17 Arguing against Fin, and losing15:36 Under three minutes, by hand, at any hour18:02 Common Room, and how leads get attributed19:07 Otterly and buyers who never see the website19:59 Fifty prompts, and building content to show up against them21:17 Narrowing the ICP after being "Stockpress for everyone"23:38 Writing with AI, and the one thing she won't hand over27:22 Claude Design for wireframes, straight to the dev team28:46 AI in customer success, and the MCP server in progress35:18 Her Friday night with no output38:37 Rapid Fire FiveConnect with Jessica Storry on LinkedIn, or learn more about Stockpress at stockpress.co.This episode is sponsored by Common Room (https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.To hear more conversations like this one, subscribe to The GOATS of Growth on Spotify, Apple Podcasts, or YouTube.

Eric Herzog says AI cut his team's content development time in half — and that anyone treating it as more than a tool is about to embarrass themselves in front of their biggest customers. A farmer still drives the tractor. A lumberjack still runs the chainsaw. What's changed is that the tool is now fast enough to make a sloppy company look sloppy at scale.Eric is CMO at Infinidat, the enterprise storage company behind InfiniBox, recently acquired by Lenovo and now operating semi-independently inside its Infrastructure Solutions Group. He runs a 15-person team across the full breadth of marketing, selling high-end systems to the Global Fortune 1000. Before Infinidat he held CMO and VP roles at IBM Storage and EMC, and helped build several startups that were acquired.His operating motto is trust but verify. Humans make errors; AI hallucinates, which is the same thing. He's seen it firsthand — someone showed him an AI-generated profile of himself with the wrong university, the wrong location, and the wrong Eric Herzog. In high-end enterprise, a data sheet claiming something technically impossible tells the buyer you're a sloppy company, and in storage people lose their jobs over a bad buying decision.The payoff is real when it's controlled. Eric's team ran an AI competitive analysis for sales, and it surfaced genuine Reddit reviews from end users explaining why they ripped out a competitor's product — material the reps now carry in their back pocket. His warning on workflow design is just as concrete: get the data sources wrong at a global manufacturer and the bulldozer bolts ship to the Corolla factory.WHAT WE COVER- Why AI is transformational but still a tool — and how open-source AI can expose your IP, customer data, and GDPR, HIPAA and PII obligations- The bulldozer-and-Corolla problem: what bad AI workflow design does to a supply chain- "Trust but verify," and the AI profile that got Eric's own biography wrong- Where the ROI actually shows up: content creation and competitive intelligence- The Reddit reviews an AI tool found that Infinidat's sales team now uses- Why a human designer still beats AI at building a deck, and why AI still gets you halfway there- Build vs. buy, and why AI inside the martech stack is becoming a checkbox, like a spare tire- Rapid Fire: the collegiate wrestler who wants to be a CEO, and why Frodo is his model for quiet leadershipCHAPTERS0:00 Intro2:18 Infinidat, Lenovo, and a 15-person marketing team5:45 What AI changed — and why it's still just a tool7:08 How open-source AI exposes your IP9:34 Bulldozer parts at the Corolla factory11:43 Trust but verify — and the wrong Eric Herzog13:14 Where the ROI shows up: content and competitive15:39 Verification still cuts the time in half17:22 Workflow automation and the AI-guided website19:02 Why the designer still beats AI on slides21:24 Playing to your audience: CIOs and King Oberon23:24 Build vs. buy25:46 AI as a checkbox: the spare-tire argument28:53 Rapid Fire Five37:18 How to reach EricConnect with Eric Herzog on LinkedIn, or learn more about Infinidat at infinidat.com.This episode is sponsored by [Common Room](https://www.commonroom.io/?utm_source=goat-podcast&utm_medium=partner), the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.To hear more on what AI and agents are actually doing inside real go-to-market organizations, subscribe to The GOATS of Growth on Apple Podcasts, Spotify, or YouTube.

Lance Hodgson thinks a huge slice of the economy — construction, oil and gas, mining, manufacturing — is about to skip SaaS entirely and jump straight to AI. Not because these industries got faster at digitizing, but because AI removes the reason they needed the in-between step.Lance is CRO at Sitemate (formerly Construction Cloud), the AI platform for what he calls "the built world." He joined as the company's first go-to-market hire and worked every seat on the way up — marketing, VP revenue, GM of Oceania, and now CRO — including years running the sales funnel himself after his own "negative perception of sales" wore off. Before Sitemate he was in marketing at Sony Pictures and CBS.Internally, Sitemate runs four categories of AI agents — workflow, intelligence, coaching, and collateral — built by Lance's chief of staff, who does the building while he defines the go-to-market logic. The team now ships three to four new agents a week. The intelligence agents are the ones he can't function without: instead of combing through call recordings by hand, agents now surface who's following the playbook — turning "high viz" from an aspiration into something that actually runs.The clearest signal of AI's impact isn't GTM headcount — Sitemate is still hiring SDRs, AEs, and CSMs at close to the same pace. It's ops. RevOps hiring has slowed sharply, because a chief of staff plus a few agents now does work that used to need a growing team under him. Engineering keeps hiring aggressively — the agents just make them ship faster.WHAT WE COVERWhy the built world (construction, oil & gas, mining, manufacturing) might skip straight to AI, bypassing the SaaS step most industries went throughSitemate's four categories of agents — workflow, intelligence, coaching, collateral — and which one Lance can't live withoutHow a chief of staff, not a platform, became the unlock for shipping 3–4 new agents a weekWhere AI has and hasn't touched headcount: flat in GTM, down in RevOps, still climbing in engineeringLance's unconventional path to CRO — Sitemate's first go-to-market hire, who'd never carried an AE quotaWhy "everyone can build agents now" and the real bottleneck is deciding which ones and what context to feed themCHAPTERS0:00 Intro1:43 Sitemate and the built world, defined3:51 Skipping straight from paper to AI5:13 Why Lance bet on Sitemate: the TAM math9:42 From marketer to accidental salesperson10:53 Building the systems: the VP of Revenue years13:36 Redefining what sales actually is20:28 Rolling out AI: the four agent categories22:31 Documentation first: mapping the funnel before building agents29:38 Intelligence agents: "High Viz on steroids"32:43 The chief of staff model: shipping 3-4 agents a week38:23 "Everyone can build agents" — the real bottleneck40:50 Headcount: steady in GTM, thinner in ops44:44 Rapid Fire Five48:33 Bonus question: lessons from "failures," not perfectionConnect with Lance Hodgson on LinkedIn (linkedin.com/in/lancehodgson), or learn more about Sitemate at sitemate.com.This episode is sponsored by Common Room, the AI GTM Intelligence Platform trusted by teams like Okta, Anthropic, and Fireworks AI.To hear more on what AI and agents are actually doing inside real go-to-market organizations, subscribe to The GOATS of Growth on Apple Podcasts, Spotify, or YouTube.

Ben Keighran got frustrated that Claude couldn't search LinkedIn. So he built a dataset larger than LinkedIn's and pointed AI at it.It's over a billion people, 10 billion records, 75 million updates a day. Not scraped — crawled from public profiles, sourced through 50 data partnerships, enriched with inference tokens. Available inside Claude and your own agents for eight dollars.Ben is a four-time founder whose products have reached over a billion people. He started coding at 13 in Australia, moved to Silicon Valley in 2006, built Chomp — the app search engine Apple acquired and turned into App Store search — then Caffeine, a live streaming platform that grew to 61 million users. Super Carl is his fourth.But the database isn't the moat, and he says so. He expects everyone to have one eventually. The thing he thinks is durable is the relationship intelligence layer on top: you upload your own contact exports, Super Carl works out who you actually know among the matches, and the AI tells you who to ask for the intro. Not scraped, not read — metadata only, and you can export or delete it whenever.Jay uses it to run his own recruiting agent. Searches that used to take weeks now finish in twenty minutes, which is why he spends part of this episode complaining that eighteen minutes feels slow.WHAT WE COVER- How Ben picks problems: watch a medium shift, then ask who won the last one and why they probably won't win this one- Why Chomp existed — apps had no metadata, so you had to type "Facebook" to find Facebook- The Instagram App Store chart hack, and what it teaches about distribution timing- Build the product first, then build the business — and how you know you've crossed over- The three things blocking AI from your network: walled gardens, disconnected graphs, and cold outreach getting worse- Why the billion-record dataset is a diminishing advantage- What "your data stays yours" actually means in practice- Rapid Fire 5: play-based learning, Steve Jobs, and the skateboard injuriesCHAPTERS00:00 The show is looking for a sponsor00:27 Chomp, Caffeine, and a billion people02:36 How Ben decides what to build06:06 You had to type "Facebook" to find Facebook06:50 Finding the first users09:46 The Instagram App Store chart hack11:07 Why Claude couldn't search LinkedIn16:39 Walled gardens, broken graphs, dead cold outreach18:04 Building a data set bigger than LinkedIn's20:41 The relationship intelligence layer22:35 Why the database is a diminishing advantage26:46 Product fit first, then build the business31:02 Rapid Fire 536:01 Bonus question: what's your biggest learn?39:10 Where the name Super Carl came fromFind Ben at supercarl.ai — free account in ten seconds, or book a meeting with him directly from the homepage.This episode is unsponsored. If you want to reach GTM leaders and founders who ignore your emails, reach out: [email protected] wherever you listen, and tell someone who'd like it.