Posts

What happens when a startup employee leaves on a Monday?

In a twenty-person engineering team, one resignation is a 5% headcount loss. The remaining nineteen absorb the work.

In an AI-pilled three-person team running twenty autonomous agents, one resignation is a 33% headcount loss.

The agents do not resign. They keep generating, reviewing, testing, and deploying. But one-third of the institutional memory that trains, prompts, validates, and debugs the agent fleet walks out the door.

One search ad every 40 minutes pays for a trillion-parameter model. One content ad every 3 minutes does the same. The math on ad-supported AI is better than you think.

When Anthropic pulled Claude Code from the $20 plan last month, it signaled an industry assumption : frontier intelligence requires frontier pricing. For open models, the economics flip.

A B200 GPU costs $4.50/hour on spot markets.¹ Google Search ads generate $38.40 CPM (cost per thousand impressions),² while Google Display runs $3.12.³

I remember growing up reading Sports Illustrated. There was a small column called “This Week’s Sign the Apocalypse Is Upon Us.”

With all the dire predictions about AI, it’s important to also spend time recognizing the tremendous pace of innovation & the impact AI is having broadly. Here’s what it fixed, taught, and discovered in the last two weeks.

In medicine

• Mayo Clinic validated an AI that detects pancreatic cancer up to 3 years before diagnosis on routine CT scans.¹
• J&J reported AI is halving the time to generate new drug development leads and cut clinical trial report preparation from 700 hours to about 15 minutes.²

In classrooms

Google Cloud grew 63% year-over-year in Q1 2026. Amazon Web Services posted 28%. Microsoft Azure hit 40%. All three are exceptional. Only one hit 63%.

The divergence is striking. AWS & Azure resell compute. Google bundles compute with its own models. Whether that explains the full gap is unclear, but the structural advantage is not : Google owns Gemini & TPUs top to bottom, with no licensing fees to OpenAI or Anthropic. Its growth may be more profitable too.

Your sales team is starting to close some terrific accounts. As your startup grows, your sales team will experiment with different sales techniques. For example, qualification, pricing, positioning, incentives and contract structure. This is a wonderful phase for a startup. However, there’s a common mistake to avoid.

Your VP of Finance should model the impact and approve each experiment. Many startups don’t do this at the early stages of go to market. Failing to consider the cash flow impacts can balloon your burn.

The inference market is the fastest growing market in the world & it’s splitting up. Each modality is developing its own inference stack.

NVIDIA’s data center revenue was flat through 2022. Then ChatGPT launched. Three years later : 17x growth.¹

Databases did the same thing. What started as one market fragmented into relational, document, key-value, graph, time series, vector, & others. Each category reflects different workload requirements : real-time transactions vs batch analytics, ACID compliance vs eventual consistency.

Recently, I wrote about customer/revenue operations, an idea that seems to be taking hold at many different SaaS companies. Instead of optimizing the performance of each individual step of the customer lifecycle, customer operations optimizes it over the entire journey. This is a fundamental change in the way a business manages its customers, and it’s now starting to be reflected in the organizational structure of SaaS startups

Two advances in thinking have led to the idea of customer operations. First, customer journeys are not linear. The classic funnel that we all picture in our minds isn’t the way that customers buy anymore. A customer may register for a SaaS product’s free trial, try it, contact customer support and decide not to buy at the expiration of the trial. But they return 90 days later and sign a contract for $90,000. Or a buyer might attend a few events hosted by a vendor, receive marketing emails over the course of 18 months, and then finally convert with an inbound call. Neither of these customer journeys is exactly the “prospect to qualified lead to contract” journey that has been canonicalized in Predictable Revenue.

The Ideal Customer Profile. The perfect customer. Can you describe it for your startup? The more precisely you can describe it, the better. That will simplify disqualification. But articulating the ICP well isn’t enough.

Vague ICPs are problematic. The company will focus on too broad a customer base, waste time and effort with unqualified prospects, and blunt their sales pitch with irrelevant value propositions.

Clear ICPs can also be problematic. To describe the ideal customer well is not enough. Let’s use an example. A clear but useless ICP might be: a disenchanted thirtysomething mechanic who likes to play German board games, read Nietzsche and watch MMA. I can clearly articulate my ICP to myself and others. The target market is clear (and niche!)

The old sales motion asked : what’s your software budget for this category?

The new motion asks customers three questions :

• What’s your software budget?
• What’s your total labor budget?
• What do you want that ratio to be in three years?

That third question shifts a software sale into a strategic planning conversation, the same conversation every board is having right now.

The higher the ratio today, the larger the opportunity. Sales runs 10:1. Support runs 4:1. Engineering runs as high as 25:1. If AI collapses the labor side, the software budget isn’t the ceiling. It’s the floor.

NVIDIA’s latest GPU rental prices on the Ornn Compute Price Index hit $4.95 per hour this week, up from $2.31 in early March : a 114% surge in six weeks.¹

The price spread over prior-generation chips doubled from $0.28 to $1.80 per hour. The new chip is NVIDIA’s B200 (Blackwell); the prior generation is the H200 (Hopper).