universal data capturer

an unstructured data logger that accepts video, clicks, audio, text, or messages to a bot. the key use case: seamlessly gather data on anything — sleep patterns, yawn counts, caffeine intake — and easily wire it to google sheets.

the interesting extension is an LLM-powered dynamic schema system: describe what kind of data you want to track to an LLM and it automatically updates your google sheet. one day you decide to track music practice time, you tell a discord bot, and a new column appears. the bot interface would support insert row, insert column, and delete column operations, with the LLM parsing natural language into these structured commands.

this is the software-side infrastructure for the broader sensory capture vision. it also connects to the universal habits system which would need data inputs to know when to trigger reminders (e.g., "button next to my bed" to log wake-up time). the personal IFTTT idea overlaps in the automation layer.

spreadsheet evaluation: originality 7/10, excitement 7/10, MVP 1-2 weeks. competitive landscape: GPT for Sheets (SheetAI) and Gemini in Sheets do AI-in-spreadsheet but with pre-defined schemas, n8n workflows can parse text to Sheets but require setup — no product dynamically creates new columns as new data types emerge from unstructured input. the LLM-creates-columns part is the novel bit. MVP could be a Telegram bot to Sheets pipeline with GPT parsing; API costs scale with usage. tech depth 5/10, labeled [WRONG FIT] ("interesting but more product than hard problem").

timeline

• [2025-12-22] captured in ideaflow — discord bot architecture, dynamic google sheets schema
• [2025-12-22] also in apple notes potential-projects — strikethrough indicating moved to ideaflow
• [2026-04-10] google sheets evaluation — 7/10 originality, 7/10 excitement, MVP 1-2 weeks, [WRONG FIT] tier