TLDR Founders 2026-04-13

Metronome CEO Scott Woody led Dropbox's Growth & Monetization engineering team, helping scale the company to 10M+ paying customers. Now he runs the billing platform behind AI leaders like Anthropic.

On this episode of Unpack Pricing, Scott explains why seat-based pricing is collapsing under AI, why customers and companies are both demanding the shift to usage-based models, and what he means when he says founders who nail the product but ignore pricing are "running a charity."

If you're building an AI product, this is worth 45 minutes of your time.

Listen to the full episode →

Competitive moats change. The moat used to be who could ship the fastest. It is now learning speed: how quickly an organization can absorb what AI makes newly possible and restructure around it. The companies that will pull ahead are the ones that can build as if they were designing their organizations from scratch, knowing what AI can do.

For a decade, SaaS companies added stock-based comp back on the way to "adjusted" EBITDA and pretended that 5-8% annual dilution was a footnote. Employees never bought it. They sell RSUs on vest day, budget the income like salary, and experience a stock decline as a pay cut. Now AI is repricing the entire SaaS universe, and companies that rode the add-back face an ugly choice: grant more shares to keep talent (brutal dilution when the stock recovers) or hold the line and watch people leave.

Product teams compound every sprint, but marketing blows up and starts over every time a new CMO or agency walks in. The ICP gets rewritten, the positioning deck gets shelved, and four months of messaging work vanishes. AI makes it possible to finally close that gap. One marketing leader with the right setup can build a living system where content draws from validated frameworks, new hires inherit full strategic context on day one, and this quarter's learnings actually feed into next quarter's campaigns instead of being forgotten.

AI now initiates most sales conversations, shifting the buyer journey. Companies are reorganizing their go-to-market strategies to adapt to AI's influence, with marketing teams building tools in-house to drive growth. While AI handles small-value purchases autonomously, complex enterprise decisions remain human-led, necessitating equipping AI as effectively as internal teams.

Nothing in your brand gets used more often or for longer than your name. It shows up on every invoice, every pitch deck, and every email signature. Most founders still pick theirs by committee brainstorm or by whatever .com happens to be available. This post walks through what separates names that stick from names that get changed in two years, and why "safe" names that try to please everyone end up saying nothing at all.

Deploy and manage AI agents to analyze and trade on prediction markets.

Conduct AI-led interviews and rank candidates based on performance instead of resumes.

Real-time database and backend built for AI-coded apps.

Building a feature has never been cheaper. Carrying one is just as expensive as it's ever been. Every addition makes the product harder to learn, harder to maintain, and harder to evolve, and products rarely fail for lack of features. They fail because they accumulated too many that nobody had the discipline to kill. When execution gets cheap, the right response isn't to ship more, it's to raise the bar for what gets to stay.

For the last decade, progress in AI came from making a single model bigger and feeding it more data. This post argues that the next leap won't come from scaling further. Current models are already good enough. The real bottleneck is how we organize and coordinate them. Better systems wrapping existing models will outperform one monolithic model trained on everything, and the field is only starting to figure out what that looks like.

Out of ~15,000 seed-funded startups per fund cycle, maybe one produces a $10B+ revenue company. The people who build those aren't a better version of great founders, they're a different species entirely. A 2 sigma founder builds a $10-100M business. A 3 sigma founder builds a unicorn. A 4 sigma founder doesn't adjust their ambition to match reality, they adjust reality. Each level is a 10x jump, not incremental.

Stop borrowing someone else's compass.

Enterprises making an AI architecture decision today are deciding who gets to own the state that builds around it.

By the time a startup needs a second product, it's perfectly set up to fail at building one.

A non-technical founder went from a few college CS classes to shipping production features daily with AI coding tools, now ranked top 1% of engineers at small startups.