Data Leads Future

By this August, our team had been writing code with AI for more than half a year. The biggest thing I noticed during this time was that our development speed went up, but code quality stayed pretty rough. Development often took just half a day, but huntin

IntroductionLast week, I burned through 7 days' worth of quota in a single day while using the Kimi K3 model on Kimi's Allegreto plan.Kimi K3 is genuinely great. Its performance is on par with Claude Fable 5 and GPT 5.6 SOL, so I ended up going full throt

This article was last updated on July 27, 2026.Today I want to talk about how I built an agentic coding loop inside OpenCode. People call this Loop Engineering.To show you how well this coding loop holds up, I used DeepSeek-V4-Pro and DeepSeek-V4-Flash fo

People are arguing again about whether prompts should use HTML or Markdown. Last year, the debate was Markdown vs. JSON.But to me, the answer has nothing to do with the format itself. Markdown, JSON, HTML, they all work well for the same reason: structure

IntroductionI gotta say, DeepSeek-V4 is cheap, works great, and has a long context window. It's already my go-to coding model. The new GLM-5.2 model is also pretty impressive. But as of mid-June, the multimodal versions of these models still haven't been