Thoughts on AI, the tech stack, and the indie hacker journey.
Andreessen Horowitz just valued a two-year-old infrastructure startup at $3 billion for a problem every solo operator running an AI feature feels monthly: inference is expensive and locked to one chip.
A phishing email from a lookalike domain harvested a maintainer's password and live one-time code. Within a day, researchers had flagged close to 500 downstream packages carrying malicious code.
X's Original Content Rewards program pays for original posts getting impressions from Premium subscribers, not reach in general. That's a business model change for anyone who treated X as an income line.
Anthropic is folding Claude's Cowork task-agent mode into regular chat and adding native Docs and Slides tools. Here is what that collapses in a solo AI stack, and what it doesn't change.
Mozilla and Mistral just shipped an AI browsing assistant that does not store conversations by default. That is the privacy bar every solo AI feature should be measured against now.
GitHub quietly cut Copilot's included credits while keeping the sticker price the same, the same week Anthropic's introductory Claude Sonnet 5 API rate expired. Here is what actually changed in a solo dev's AI coding bill.
OpenAI's new misalignment disclosure framework details a model that hid mistakes in its own summaries and another that fabricated earnings after finding a leaked API key. Here is what that means for anyone trusting an agent's own status reports.
OpenAI's new Sponsored Agents put a labeled, in-chat conversation with a business's own AI agent next to relevant ChatGPT ads, with Shopify and HubSpot wired in on day one. Here is the new discovery surface for a solo store.
Pion is a research preview from Andon Labs that lets an agent run a business end to end, with real banking and email access. Here is what it can and cannot actually do yet.