First Harmonic

Such an awesome evening with the team at First Harmonic — thank you Ali Rowghani / Michael Tan / Mia Mabanta!

This week, First Harmonic hosted its first sales-focused Builders Table (Sellers Table?) with top AEs from Decagon, Rippling, fal, Serval, Hex, Stripe, CodeRabbit, Figma, and other great companies. Some interesting takeaways from the room: 1/ Most sellers are using AI for meeting prep and follow-up, not as much for building decks or writing emails (!). 2/ The thing sellers most WISH AI could do for them is automate CRM input (hard relate). 3/ Most sellers think that there will be fewer SDRs in five years, and that they'll have different skills than they do today. 4/ Most sellers think Salesforce will eventually get displaced, but not for 10+ years. 5/ The "AI tool" that sellers value most isn't an AI-native tool at all; it's Gong. Personally, I learned that the most cracked sellers are the ones who show up at a dinner on the last day of the quarter, the week before 4th of July! Thanks to all who joined us. We want to do more of these dinners: for engineers, salespeople, marketers, you name it. DM or email me if you want to get on the list.

Today we’re announcing Unify’s next chapter. 3 years ago, Connor and I founded Unify ~2 months after ChatGPT’s release, with the deep belief that AI would change how sales works. It’s the reason why we joined OpenAI’s first Converge Accelerator program. The long-term promise of AI in sales was there. But the models weren’t good enough to execute anything besides basic email copy. The 2024/2025 Unify was designed for the era of humans clicking buttons. Over the last 6 months, that world has flipped upside down. Today, we’re announcing: (1) Unify is now like Claude for outbound sellers. (2) AEs & SDRs can now be GTM engineers (40+ data sources available including new ones like Store Leads, Openmart, Adyntel, The Swarm and more). (3) We have self-service plans starting at $20/mo. Initial traction has been promising. Since launching a private beta a few weeks ago, we’ve had 57,548 queries completed, growing 45% w/w with daily active usage from companies like Pylon, Juicebox and Console. Our new direction is one that's focused on sellers. And it builds on top of the foundations and primitives that we've built over the last 3 years. Try the new Unify here: https://lnkd.in/ghs9-MfN

Forge is now Poetic. We're announcing our customer impact and also announcing our Series A backed by Kleiner Perkins, Founders Fund, OpenAI, First Harmonic, Genius Ventures, and other top investors. We believe every business will be re-founded as a software business. To get there, the rules that run a business, most of which live in people's heads, must become software. AI can't improve what it can't touch. We’re building an AI platform that learns every edge case, encodes knowledge into code, and runs processes deterministically—fully auditable, with 99%+ accuracy on the world’s most complex processes. We combine the intelligence of AI with the guarantees of software. Poetic has been deployed in production globally, driving double-digit millions in savings for Fortune 500s, over 500,000 disputes handled at Chime, and 100% procedure adherence at SoFi. The Executive Chairman of AIG, Peter Zaffino — “Poetic has the potential to redefine how companies manage complex workflows. Using sophisticated technology and process reengineering, they have achieved 99%+ quality outcomes on multi-hour processes - delivering real enterprise value.” Soon, every business will be software, and the people who write that software will define what the next generation of companies looks like. Come build in the new paradigm and re-found the biggest industries in the world. News coverage linked in the comments.

One of my favorite evenings in recent memory. We hosted 40 of SF’s best engineers at our Builders Table dinner with Boris Cherny, Mike Krieger, and members of the Claude Code team. We took a quick poll in the middle of dinner on some topics, and we had some interesting results: 1) It turns out that out of all the agentic coding tools out there, most folks in the room (to no surprise) trusted Claude the most 🤖 2) The next application layer unlock will come from persistent memory and better context ingestion. 3) The room was pretty split on whether there would be more or fewer software engineers in the world in 5 years. Thanks to the Anthropic team (Kamal Ndousse, Omid Mogasemi, James Baskerville, Alistair Smith, Robin Jayaswal, and especially Raza Habib), and to everyone who spent their evening with us. If you're an engineer or researcher interested in a future First Harmonic dinner, DM me.

We hosted a Builders Table event with leaders from Anthropic last week. It was amazing to hear about their experiences using Mythos internally, managing the unprecedented growth they are seeing, and learning about new products they are hoping to build. Thanks to Raza Habib, Mike Krieger, Boris Cherny, Kamal Ndousse, Omid Mogasemi, James Baskerville, Alistair Smith and Robin Jayaswal for joining us along with 40 of the Bay Area's best engineers.

This was maybe my favorite Builders Table yet. We hosted Boris Cherny, Mike Krieger, and members of the Claude Code team to talk about how coding will evolve and what it means for the day-to-day experience of engineers. 40 of SF’s top engineers came. Some results from our mid-dinner poll: 1/ The biggest coding bottleneck is verification, not model capability. 2/ The next application layer unlock will come from persistent memory and better context ingestion. 3/ The jury is out on whether there will be more or fewer engineering jobs in five years, but for now the coding "tool" people trust most after Claude is... themselves. Keep thinking indeed. Thanks so much to the Anthropic team (Kamal Ndousse, Omid Mogasemi, James Baskerville, Alistair Smith, Robin Jayaswal, and especially Raza Habib), and to everyone who spent their evening with us. We're hosting another one of these next month, on the topic of AI safety. If you're an engineer or researcher and want to come, DM me!

📣 We just raised a $45M Series B to transform how every company sells. In the next few years, every winning sales team will be powered by a system that continuously thinks about every account and how to progress it down the funnel, guides every rep and leader on what to do next, and helps them to do it. The greatest privilege has been getting to work with and learn from many of the most forward-thinking enterprises out there, like Samsara Ramp Ironclad Attentive — companies with billions in revenue and thousands of sellers, but who are living in the future and recognize the immense shift that is occurring — a few of whom you can hear from in the video below. And this is only the beginning. As Anshul puts it (my favorite line in the whole video!) — we're excited to continue building at the frontier of AI to make customers incredibly happy, incredibly successful, and pull them into the future that we know is inevitable. Intelligence-Led Revenue is here. Thanks to Sofia Chierchio for the feature in Forbes (link in comments), to our new and returning investors at TCV First Harmonic Bain Capital Ventures (BCV) First Round Capital Alkeon Capital, our team, and, most importantly, our customers!

Last week, we hosted 20 of the sharpest researchers in post-training from companies like Applied Compute, OpenAI, Cursor, Thinking Machines Lab, Mercor, Cognition, xAI, Flapping Airplanes, Standard Intelligence, and others to discuss where the field is headed. Here's what kept coming up: 1/ Reward modeling is a bottleneck. We can train models to carry out increasingly complex reasoning, and it's easy to check whether they get the right answer. The key challenge is judging and rewarding the reasoning process accordingly, which is much harder to do with sparse, human preference reward modeling at scale. 2/ Evaluation as a whole is still artisanal. Everyone's building their own eval stack, and nobody loves theirs. The tooling layer for evaluation is wide open. There are many forums for founders to learn from each other, but far fewer for the engineers and researchers thinking about these challenges. If you're interested in attending a future First Harmonic dinner, DM me.