LLMs make one-off webapps viable: OCR proofreading and a decade of Tumblr analysis
StewartalsopIII · x · 2026-09-19
A workflow shift enabled by LLMs: instead of building universal platforms, spin up lightweight one-off webapps for each specific project. Two examples:
- Proofreading an entire book's OCR (very difficult scans, line by line): a small app put the OCR output, color-coded confidence, and the original scanned line in one table for fast reading and fixing
- Categorizing 11 years of Tumblr history for later analysis of aesthetic preference trends
A textbook case of "software one-offs": when generation cost approaches zero, tools can be scoped to a single task and thrown away.
More from coding & agent
- Hamel Husain Explains What a Trace Is in LLM Evaluation — HamelHusain · 2026-09-19
- miro-mcp-server lets AI assistants control Miro whiteboards — modelcontextprotocol · 2026-09-19
- Automating App Store Connect Submissions with AI and asc-cli — rudrank · 2026-09-19
- Notion Rebuilt Collaborative Editing on CRDTs to Stop Losing Concurrent Edits — kevinakwok · 2026-09-19
- SurrealDB Ships Docs Skill for Coding Agents and Agent Memory in Studio — mattturck · 2026-09-19
- Smaller models aren't always cheaper: hidden review and rework costs — zeuslac · 2026-09-19