YC is quietly building the infrastructure for the agent economy while Google demos party tricks. The gap between what's powering real AI businesses and what's getting the headlines keeps growing.
01
YC President endorses Exa as the search engine for AI agents
Garry Tan called Exa "what I trust for all my agents" and said Y Combinator uses it across their OpenClaw and Hermes agent systems. He described it as faster, more reliable, and more complete than alternatives for web search in AI applications.
Why it matters: When the most influential startup accelerator in Silicon Valley standardizes on one search tool for agents, that's a signal about which infrastructure will power the next wave of AI companies. Every YC startup building agents now has a default recommendation.
Google's Project Genie lets you build games in minutes
Google Labs showcased Project Genie, which generates playable games from simple prompts. Users choose characters and settings, then Genie creates the complete game experience. The tool is being demonstrated at Google I/O's sandbox this year.
Why it matters: This is Google's answer to "what happens when AI can create interactive experiences, not just content?" If you can build a game as easily as writing an email, the app store is about to get very crowded.
Vercel CEO claims new AI feature will reach "42% of the web"
Vercel CEO Guillermo Rauch announced a new capability that will support "every model, every provider, every modality" including text, image, video, and audio. He claimed it will bring AI to 42% of websites, referencing Vercel's market share among web frameworks.
Why it matters: Vercel powers millions of websites through Next.js and other tools. If they're embedding AI capabilities directly into their platform, developers won't need to integrate OpenAI or Anthropic APIs separately. That's a much simpler path to AI-powered websites.
AI entrepreneur connects model improvements to revenue spikes
AI builder Swyx noted a direct correlation between AI model performance improvements and revenue at "Agent Labs" companies, referencing Sam Altman's advice to "build a business that gets better when models get better." He pointed to a discontinuity in Q4 2025 when model capabilities jumped.
Why it matters: This is the first concrete data point showing that AI model improvements directly translate to business results for companies building on top of them. If you're building AI products, your revenue might be more tied to OpenAI's release schedule than your own product decisions.
Product manager Peter Yang suggested that avoiding companies with frequent layoffs and performance management cycles might be better for mental health, responding to ongoing discussions about tech industry work culture.