AI & the future
What AI actually changes for a small business online, minus the hype, and what's just noise.
Writing web content that AI search engines will actually cite
More people ask an AI assistant instead of scrolling ten blue links. Here's how to write pages those tools quote instead of skip.
New AI models ship, and one big one slips
OpenAI and Anthropic rolled out new flagship models this summer while Google delayed its top Gemini after it fell short on coding and reasoning.
The EU AI Act's real deadline arrives
On August 2, 2026 the EU AI Act becomes largely applicable, with transparency rules for chatbots and AI-generated content, and real fines behind them.
Claude Opus 4.8 runs codebase-scale migrations
Anthropic's Opus 4.8 catches roughly four times more code flaws than its predecessor and can run hundreds of parallel subagents on very large jobs.
Gemini 3 lands, and the bar moves again
Google shipped Gemini 3, its most capable model yet, topping reasoning leaderboards with a million-token context window and strong coding scores.
Olmo 3 ships as a truly open AI model
The Allen Institute released Olmo 3 with its full training data, checkpoints, and pipeline, not just downloadable weights.
OpenAI ships its first open models since GPT-2
OpenAI released gpt-oss, two downloadable open-weight models under Apache 2.0, its first open release in roughly six years.
OpenAI ships GPT-4.1 with a million-token window
OpenAI released GPT-4.1 and its mini and nano siblings through the API, with big coding gains and a one-million-token context window.
OpenAI's Operator tries running your browser
OpenAI released a research preview of an agent that clicks through websites on your behalf to book, shop, and fill forms.
DeepSeek R1 shakes up the AI cost math
A lab released an open-weight reasoning model rivaling top systems at a fraction of the training cost, rattling markets.
Next step
Reading is good. Building is better.
Got something you're dreaming up? Tell us about it and we'll map a course, no pressure.