BitsExplorer finds what people are building with AI, every day. See today's picks →
TOD reads text and images, extracts facts, and chooses the next action. Explore measured support benchmarks, confidence-based handoff, and flat per-decision pricing.
Description excerpted from the original listing, which is linked below.
This page shows all the answers I scored for my project on PSA reports. 172 answers on the report files and 88 on the photos, from Qwen3-VL-4B and MedGemma 1.5 4B. For each one you see the report image, what I expected, the full answer of the model and my scores. You can filter by model.
Concept redesign of the Paint The World Foundation website (Nuxt 4 + GSAP), by BaseTech
Convert images to the wplace palette with OKLab color matching and Bayer/AB lattice dithering
Stock vs fine-tuned PP-OCRv6 on receipt photos
Agent Skills for Claude Code and other agents: design-inspiration looks at real design galleries before building; copywriter writes copy from facts and edits out the AI tells.
A ShaderToy-style playground for the SNES PPU. Write Lua, watch the frame update live, and publish tiny graphics toys.
MUON is the governed harness above coding agents. Direct Claude Code, Codex, Cursor and OpenCode as one crew, on your machine, with shared memory you confirm and gates you control.
What you want tested. Run by browser agents. Signed off by you.
Every Monday: how many projects were built with AI that week, which tools built them, where they came from, and the week's picks.
Read the latest week