Summarize a YouTuber's latest video into a Google Doc daily
A Make scenario runs at 9 AM, finds a YouTuber's newest video, has Hugging Face summarize it, and appends the summary to a Google Doc.
Built five Claude Code agents (cleaner, analyzer, visualizer, report writer, summarizer) chained by an orchestrator for end-to-end data analysis.
Built five Claude Code agents (cleaner, analyzer, visualizer, report writer, summarizer) chained by an orchestrator for end-to-end data analysis.
Bronson, a data analyst, used an AI app-idea agent to produce a 713-line specification. He then pointed Claude Code at it. Over about five hours Claude Code built five agents plus a @full-analysis orchestrator that passes files between them. It wrote its own tests and fixed its own errors. He only approved permissions, about 20 times.
The build produced 627 passing tests and cost about $35 in estimated API usage. It has not yet been run on real work data, and outputs still need analyst review.
The spec was the hard part. Pro plan is enough, though usage limits slow you down; he moved to Max for convenience.
Set up a chain of subagents in this project for analyzing [dataset file]: a cleaner, an analyzer, a visualizer, a report writer, and a summarizer, run in order by an orchestrator. Have them clean the data, find the main patterns, make [number] charts, write a full report, and finish with a one-page executive summary. Save each stage's output in its own folder and keep the original data unchanged.
A Make scenario runs at 9 AM, finds a YouTuber's newest video, has Hugging Face summarize it, and appends the summary to a Google Doc.
An n8n workflow scrapes Reddit posts and comments via Bright Data, extracts concerns and themes with Gemini, and writes reports to Notion.
Plan, execute and review a cross-cutting refactor in small batches, with Claude Code making the edits and GitHub Copilot reviewing each PR.