AI Editor Wars: Windsurf vs. Cursor, Anthropic's Power Play, and The "Gladys" Experiment

This week on Fresh From the Labs, Shilpa, Kevin, and Jared dive into the swirling drama and strategic plays shaping the AI landscape. First up, the Windsurf and Anthropic saga: with rumors of an OpenAI acquisition of Windsurf, Anthropic has reportedly given Windsurf just five days to remove all Anthropic models. We unpack the implications of this aggressive move, from competitive intelligence concerns (is OpenAI training on Anthropic's models via Windsurf?) to the potential for a fragmented AI editor market where model access becomes a key differentiator, potentially harming developers and open source.

The conversation then shifts to the evolving user experience (UX) for AI products. Kevin shares his experiences building a voice-first email agent, highlighting the fine line between a magical, futuristic experience and a frustrating one, and the surprisingly good performance of OpenAI's latest real-time voice API. Jared introduces his "Gladys pattern" experiment – building AI systems where agents think they are emailing each other to manage tasks. This explores a UX beyond chat, aiming for "ambient agents" that work implicitly, and we discuss the fascinating (and flowery) system prompts that bring these agentic personalities to life, plus the challenges of evaluating such behavioral systems.

Finally, we touch on OpenAI's new meeting recording and summarization functionality built directly into ChatGPT. While a convenient feature for users, it underscores OpenAI's relentless push into product territory, raising questions about the future of standalone SaaS tools as AI platforms consolidate interfaces and subsume niche functionalities.