A Personal Data Project

Your Saved Posts Are a Diary You Never Wrote

11,323 Instagram saves. Three years of private attention. An AI that reads every post, watches every video, and publishes an editorial magazine about the person who saved them.

11,323

Posts Analyzed

516

Gaza Events

760

Recipes Extracted

225

Person Profiles

21

Sideshifts

114

Training Exercises

5

Deep Dives

What Happens When AI Reads Your Archive

Saves weigh 3x more than likes in Instagram's own algorithm — the strongest signal of genuine interest. But the feature has had three updates in nine years. No search. No export. No tools. Your saves sit behind three taps, in a folder the platform never wants you to think about.

This project takes them back. Four AI models, each doing what it's best at: Gemini watches full videos. Whisper transcribes audio. Sonnet synthesizes meaning across five text sources. Opus extracts structured data from what it sees. Not a prompt — a pipeline of agents that extract, review, and reframe like a strategist, analyst, and editor passing work between them.

97 → 760

Recipes extracted from the food collection

A boolean flag from vision analysis found 97 recipes. AI agents reading all five text sources per post found 612. AI agents that watched the cooking videos found 760. A 684% improvement — not from better data, but from better seeing. A silent video of a chef slicing fennel became a complete recipe with 7 ingredients and 8 steps. The data was always there. You just needed an AI that could watch.

Likes and comments are social behavior. Saves are private intent. This dataset is the closest thing to a real signal of curiosity, concern, taste, and identity.

Five Deep Dives

Each collection becomes its own editorial project — not a filtered dashboard, but a purpose-built narrative with its own voice, analysis, and design.

What the AI Sees

Not What Software Does

Everyone's talking about AI replacing SaaS. That's thinking too small.

A traditional pipeline is deterministic — same input, same output, same pre-built charts. It only answers questions someone already thought to ask. The interesting work has always been human: What should we ask this data? What's actually valuable here? How do we turn raw numbers into something someone would read?

This skill orchestrates agents that loop in and out of the data — extracting, reviewing, reframing — the way a strategist, analyst, and editor would pass work between them. Each step feeds the next. Each output gets reviewed before it moves on. The result isn't a dashboard. It's a publication.

Not what software does. What an agency with expensive software and technical data scientists does — made reproducible. One person built five editorial deep dives, 760 structured recipes, 225 person profiles, and a psychological portrait. The question isn't what we built. It's what you could build with the same methodology applied to your own data.

SaaS automates processes. A skill automates expertise. Are skills the new software, or the new agency?

Explore

Built by Simon Strumse with Claude as engineering partner

Next.js · Convex · Python · Claude · Gemini · Whisper · Tailwind · Vercel