TLDR Founders 2026-03-30

Most tools are still locked to their own database, blind to everything users already have in Slack, GitHub, Salesforce, Google Drive, and dozens of other apps. That's the ceiling on what you can build.WorkOS Pipes removes it. One API call connects your product to the apps your users live in. Pull context from their tools, cross-reference data across silos, power AI agents that act across services. All with fresh, managed credentials you never have to think about.Start building connected products →

190 companies presented this week. 85% were AI-first. 14 hit $1M ARR before demo day, the highest ever for any YC batch. The bigger signal here is that the bar for what counts as "early stage" keeps moving. $1M ARR used to be a Series A milestone. Now, it's table stakes for a demo day presentation.

The AI market isn't the same as the Dotcom bubble of the past. The technology is still early, and demand is pulling spending forward. The software selloff is real, but it won't hit every layer equally, as each has its own moats. Across the last few technological revolutions, the companies that became durable winners were predominantly founded in years four and five of each platform shift. We are now in that window for AI.

Every major tech shift creates a new growth channel. AI is becoming a decision-maker - they're starting to choose products. This new growth channel is embedding products directly into the AI tools that people use. This impacts strategy, product, and engineering. The companies that will be marketing to the interface layers that make decisions on users' behalf.

The hardest problem in AI right now isn't the model, it's the pricing. In traditional SaaS, you could underprice a seat, and it didn't matter because the marginal cost of serving another user is basically zero. Now every agent action, every "magic" feature triggers real inference costs on the backend. Price too low and you're literally paying customers to use your product. Price too high and you lose to the competitor who figured it out first. Per-token and credit-based pricing models are about to become the default because they're the only ones that align cost structure with value delivered.

Littlebird observes your screen and meetings, building a private memory that grows with you. No setup, no catching it up. This is what AI was supposed to feel like.

Verify Humans with Short Video.

Mistral open-sourced a text-to-speech model that outperforms ElevenLabs on benchmarks.

Practice and improve startup pitches with feedback.

Software companies have three paths right now: accelerate revenue 10%+ YoY through AI-native products, get stuck in the messy middle, or rebuild to 40-50% operating margins. Even the profitability path is deceptive. Most companies that look profitable on FCF margins are still in the messy middle once you account for stock-based compensation. The only real option is to accelerate. If you're not rewriting your product around AI, you're choosing a slow death.

Antifragility is the premise that some things, including people, improve from stressful situations. Disorder and uncertainty are the rule at startups. Over time, antifragile people will consistently improve their value/skills due to the continued exposure to a startup's disorder.

The Strait of Hormuz is the most critical energy dependency in the world, and it has no failover.

There is surprisingly little written about working with large organizations in the age of AI - this post attempts to fill that gap.

Heidi Sturrock's broad match campaign mistake unexpectedly became an advantage by turning competitor calls into sales leads.