Outside-in Brain

One click turns what your team discovers online into shared insights, actionable briefings, and an AI podcast.

Comment

Inspiration

Every company is building an internal brain. We built the missing half.

Companies are getting better at remembering what's already inside their walls: documents, meetings, Slack conversations, and decisions. But some of the most valuable knowledge comes from outside the company.

A colleague watches a brilliant conference talk. Another discovers a new research paper. Someone screenshots a slide, bookmarks a podcast, or finds a competitor doing something unexpected.

Then the tab closes. The screenshot disappears into a camera roll. The insight never reaches the people who could use it.

We wanted to solve a simple problem: What if everything your team discovers in the outside world could become shared intelligence, without giving everyone another tool they have to remember to check?

That's Outside-In Brain.

What it does

Outside-In Brain turns the things your team encounters online into a shared, actionable intelligence briefing.

1. Capture with one click. A Chrome extension lets anyone save articles, YouTube videos, podcast pages, PDFs, social posts, or highlighted text into a shared brain. You can also paste links or notes directly into the web app. Capture is intentional: nothing gets collected automatically.

2. Connect the dots. Instead of producing another pile of bookmarks or isolated AI summaries, Outside-In Brain looks across what the team has collected. It groups related discoveries, identifies the most important ideas, and produces up to five prioritized takeaways. Each takeaway includes supporting detail, links to its original sources, and a concrete "What to do with this" recommendation.

3. Turn insights into something people actually consume. Once or twice a week, the brain produces an editorial-style briefing and a conversational podcast. An AI interviewer asks the questions a curious teammate would ask, while an AI analyst connects the ideas. People can read the briefing or listen on their commute.

4. Build a collective learning habit. Every small individual discovery can contribute to what the entire team knows. Instead of requiring everyone to search a knowledge base, the most important insights come to them.

We also built a no-login interactive demo: paste in a few discoveries, click Build this week's issue, and watch them become a briefing and a two-voice conversation.

How we built it

We used a TypeScript monorepo built on App-Starter, with a Next.js/React frontend and a NestJS/Prisma backend.

The Chrome Manifest V3 extension captures selected content, page text, metadata, and YouTube captions where available. PostgreSQL stores workspace brains, saves, issues, and provenance. The application supports team workspaces, authentication, and notification preferences.

Claude is our AI editor. It synthesizes the saved material into structured takeaways, preserves source references, proposes actions, and writes a conversational podcast script. We validate structured output before publishing it.

For the full audio pipeline, we integrated text-to-speech with distinct interviewer and analyst voices, audio chapters, and playback controls. The lightweight demo can also play the conversation using browser speech synthesis.

We integrated Triggerware into the digest orchestration layer to detect when a brain is due for a new issue and initiate processing, with a conventional scheduler as a fallback.

The architecture supports Vercel for the web app, Railway for services, and Supabase/PostgreSQL for persistence.

Challenges we ran into

Capturing the messy internet. Articles, YouTube videos, highlighted passages, and PDFs all expose different kinds of content. We needed capture to feel effortless without requiring users to manually clean up everything they saved.

Turning many sources into one coherent story. Summarizing five articles independently is easy. Finding what they mean together is harder. We designed the synthesis around shared themes, priorities, supporting evidence, and practical actions.

Making AI output genuinely useful. A fluent briefing can still be wrong or vague. We focused on preserving links to the underlying sources, constraining the AI to the submitted material, and turning abstract observations into specific next steps.

Delivering a complete experience. We wanted more than a dashboard or chatbot. That meant building a browser extension, workspace model, AI editorial pipeline, audio experience, and an immediate demo that people could try without registering.

Accomplishments we're proud of

The interaction is incredibly simple: one click to save, one briefing to learn, one conversation to listen to.

We built around a different idea of company intelligence. Most knowledge tools wait for someone to ask a question. Outside-In Brain proactively delivers the ideas the team might not even know to ask about.

We're especially proud that the output isn't simply a generic summary. It connects multiple discoveries, retains their sources, and makes the next action explicit.

What we learned

The hardest part of knowledge management isn't storing knowledge. It's getting the right knowledge into the right people's attention while it still matters.

We also learned that the best AI interface isn't always another chat window. Sometimes it's a useful briefing on Monday morning or an interesting conversation on the drive home.

What's next

We want to connect Outside-In Brain to Slack, Google Drive, and existing internal company brains, so discoveries from the outside world can inform internal work automatically.

Next, we want to track how ideas evolve across successive issues, learn which insights led to actual decisions, and give agents access to the team's continuously refreshed understanding of the world.

The vision: a company brain that doesn't just remember what your company knows, but helps it discover what it needs to know next.

Built With

Share this project:

Updates

Submission history