Matt Thazhmon

I used Claude Code to book meetings with billion-dollar companies in one night. Here's the full system so you can copy it: $7 billion fund. $15 billion fund. The world's largest alternative data marketplace. All responded within 24 hours. One terminal. Claude Code. 11 PM to 3 AM. Total cost to run the entire pipeline: less than a coffee. Sounds too good to be true? I would have thought so too... But here's what most people get wrong about cold outreach: The message is the last 5%. 95% of whether someone replies is decided before you type a single word. Who are you targeting? What do they actually care about? Are they a real buyer, or three layers removed from the budget? Most outreach looks like this: → Pull a list from Sales Navigator → Write a template → Swap in first names → Blast 500 people → 2% reply rate → "Outreach doesn't work" This system does the opposite. Three stages: 🔍 Stage 1: Deep Buyer Research Before writing anything, I ran research prompts to understand how buyers talk to each other internally. Not "what do they do." How do they evaluate what I'm selling. One prompt through Perplexity. Fifteen minutes. More buyer intelligence than most sales teams produce in a month. Five buyer segments. Five completely different conversations. Same product. ⚙️ Stage 2: Enrich, Score, Filter 214 leads went into the pipeline. 30 came out as A or B. The other 184? Noise. Claude Haiku scored every lead. A/B/C/R. Structured output. The system filtered 85% before a single message got written. ✍️ Stage 3: Messages Assembled From Intelligence Claude Sonnet wrote personalized messages using enrichment data. Only for A and B leads. Never for C or R. Every message used the buyer's internal vocabulary. Not marketing language. Their language. The two-LLM trick that made it cost pennies: → Haiku handles everything fast and cheap (summaries, scoring, filtering) → Sonnet handles the one thing that needs to be good (the actual message) Cheap intelligence at volume. Expensive intelligence where it counts. Here's what happened: ☑️ Eagle Alpha's CDO booked a call same-day ☑️ Two people at a $15B fund replied independently (neither knew the other responded) ☑️ $7B fund's Global Head of Data Sourcing replied within 24 hours ☑️ Wrong ICP batch from V1: 0% reply rate. Same product. Same datasets. The wrong batch is the control group. The difference wasn't better copy. It was better targeting driven by better research. Every tool here is publicly available. No proprietary stack. The moat isn't the tools. It's sitting down at 11 PM and not standing up until 3 AM. Knowledge is free. Execution when nobody is watching is not. What does your outreach research process look like before you write the first message? 💾 Save this for your next outreach campaign. ♻️ Repost to help someone rethink how they do cold outreach.

82

8 Comments

Figma just acquired AI startup Weavy for ~$200M. AI workflows are changing how creative work scales. Here’s why founders should pay attention 👇 Weavy is a Tel Aviv–based AI image & video generation startup. This is Figma’s most important strategic move since Adobe’s attempt to buy Figma was blocked. We’ve entered the AI design consolidation era, where major platforms are becoming AI-first for design and workflow automation: → Canva acquired Affinity (to expand into pro design) → Adobe acquired Rephrase.ai (AI video generation) → Freepik acquired Magnific (AI image upscaling) Because here’s the competitive reality: • Adobe controls the production suite • Canva controls lightweight brand creation • Framer controls design-to-web automation Companies now need to own workflows end-to-end to dominate the market. That means controlling: creation + automation + distribution Because in software, public markets reward leverage. It’s no longer about who has the best design tools but who automates the canvas itself. --- Our team at FE International is more bullish than ever on the rise of AI-first businesses. If you're a founder curious about your company’s true value, get a free valuation. (Link in comments) 👇

35

9 Comments

Anthropic released Claude Opus 5.5 today, and it is cheaper than the model it replaces while scoring higher on every benchmark they published for both. That combination is rare enough to be worth writing down. $4 input and $20 output per million tokens, against $5 and $25 for Opus 5. Cache reads drop from $0.50 to $0.20, so anything replaying a large system prompt on every call saves well past the headline 20%. Two things I would check before swapping the string. Extended thinking cannot be disabled on this model, and thinking tokens bill as output, so your per-request output count will likely rise. And the widest margins are on agentic and terminal work, not chat. Terminal-Bench-Science goes from 29.0% to 58.7%. Humanity's Last Exam moves four points. Anthropic says in its own announcement that benchmark margins are a less reliable guide to real quality than they used to be. Unusual thing for a vendor to publish about its own launch, and worth taking at face value. Run your own evaluation. I wrote up the pricing table, the cost math on a realistic monthly workload, and what actually needs attention in production. https://lnkd.in/g3tXsKYF

7

1 Comment

Cursor hit $500M ARR. Lovable went from $17M to $75M in five months. On paper, AI coding tools look unstoppable. Then you check the traffic data. Lovable is down 40%. Vercel's v0 dropped 64% since May. Bolt.new fell 27% since June. When revenue climbs while usage falls, that's not growth. That's a warning sign. Three forces are working against generic AI coding tools: → Developer sentiment is cooling. Stack Overflow shows usage up from 70% to 76%, but favorability dropped from 77% to 72%. People are using the tools more and liking them less. → Enterprises are hitting compliance walls. FINRA, GDPR, HIPAA. Many tools weren't built for enterprise procurement. → Switching costs are too low. When every tool feels the same, users bounce. There's no moat in being "pretty good at everything." Meanwhile, vertical AI is scaling fast. The market sits at $12.9B in 2024, projected to hit $115.4B by 2034. Big tech is buying specialized applications, not building generic copilots. ServiceNow paid $2.85B for Moveworks. M&A hit 177 deals in Q2 2025, nearly double the quarterly average. We're in Wave 2.0. Infrastructure is maturing. Value creation is moving to the application layer. CB Insights tracked 1,500+ tech markets. The most AI deals in Q2 went to companies solving specific industry problems. LLM developers tied for ninth place. The tools that win won't be the loudest. They'll be the ones that become essential. Full piece in Forbes (link in comments)

396

103 Comments

Extremely good insights by Leon Eisen, PhD . Discussion about two common models that startup founders seems to follow. Which one IS better or is it wiser to use them both. Very valuable advices. Recommendation to all founders.

4

1 Comment

Software beyond the screen companies have an early revenue path that pure software startups rarely touch. Government agencies fund deep tech R&D through SBIR grants, up to $2 million per award, and when handled right, those grants convert into recurring revenue. Ubiquity Ventures-backed WindBorne Systems is proof! The company just announced its $37 million Series B, and it got there in part on SBIR revenue: more than ten awards across NOAA and the military, each one converted into a recurring contract with the agency. In the newest Ubiquity University module, WindBorne CEO John Dean joins me for a crash course on SBIRs for founders. A few of the hard-won lessons: 👉 Skip any grant that will never become revenue. The good ones are the start of a long-term customer relationship 👉 Sell it like enterprise sales. Find the real buyer inside the agency, understand why the solicitation exists, and build a champion before you submit 👉 Get expert advice, but never outsource the proposal. Nobody knows your tech and your story like you do 👉 You write the milestones and the payment schedule, so make sure they work for your business Full video on Ubiquity University plus the WindBorne Series B news, links in the first comment. #SoftwareBeyondTheScreen #PhysicalAI

24

3 Comments