We've been building Enlighten Copilot for a while now. Today's the first time we're showing it publicly.
It's an AI copilot for your professional network not just LinkedIn connections, but any contacts you work with. You can semantically search across everyone you know to find the right person for the right moment.
Hiring? It finds the engineers in your network with the right background. Visiting a city? It tells you who you should catch up with. Got a meeting tomorrow? It briefs you on everyone in the room.
Under the hood we're using @turbopuffer for vector search which is the same object-storage-first engine behind Cursor, Notion, and Linear. When you're searching across thousands of connections and as we scale out our users that speed and cost efficiency is going to really matter.
We've also been using @elevenlabsio for a while voice agents, dictation, TTS across the app.
So when #ElevenHacks dropped and the theme was turbopuffer + music and sound effects, our first thought was: "For a B2B app, this is going to be interesting."
So we built Network Briefing.
You search your network. Turbopuffer finds the right people. Then ElevenLabs generates a podcast-style audio briefing narration, background music, and sound effects all composed live in the browser.
The music isn't stock. The prompt is derived from what the search returns industry, seniority, result density. Every search literally sounds different.
It works incredibly well on single profiles. And for meeting prep hearing a 30-second briefing on three people before you walk in its genuinely useful.
This is our first #BuildInPublic post. More coming on how @turbopuffer powers our RAG building modern voice applications with @elevenlabsio for voice in a B2B context. Plus from a UX/UI perspective the new AI interface patterns we're pushing beyond chat and voice.
6/ If you build, manage, or rely on software your business can't function without this is the most important thing you'll read this week.
https://t.co/8gmAJ02W5a
1/ Every piece of software ever written just became a liability.
Anthropic's new Mythos model has found thousands of critical zero-day vulnerabilities across every major operating system and web browser.
One bug hid in OpenBSD for 27 years. Another survived 5 million automated tests undetected.
5/ We'll be looking at what Mythos and Glasswing mean for how we build, audit, and maintain software for our clients.
Every software company should be doing the same.
7/ For businesses adopting AI right now, this competition is the best thing that could happen to you.
Both platforms are being pushed further, faster, and in genuinely different directions.
1/ I posted about Claude Opus 4.6 this morning.
Minutes later, OpenAI dropped GPT-5.3-Codex.
Same. Day. ๐ฅ
Super Bowl ads. Same-day model drops. CEOs trading shots on X.
Here's what we actually know so far ๐งต
6/ But I'm hitting pause before declaring winners.
Both claim state-of-the-art on overlapping benchmarks. Independent testing will sort this out in the next 48 hours.
7/ We started embedding AI into how we build and deliver software two years ago. Not as a novelty as a fundamental shift in how we serve our clients.
The speed of change is unlike anything I've ever seen.
https://t.co/EcF6a1vKfX
1/ Anthropic just released Claude Opus 4.6, continuing to push their lead.
If you need proof, look at what happened this week.
The ๐ฆ๐ฎ๐ฎ๐ฆ๐ฝ๐ผ๐ฐ๐ฎ๐น๐๐ฝ๐๐ฒ a $285 billion selloff in software stocks was triggered by Anthropic's CoWork AI agent.
Thomson Reuters dropped 16% in a day. LegalZoom fell 20%.
https://t.co/k95yaNvVm1
6/ At Enlighten Designs, Claude has been our primary coding model for a long time. But CoWork with our business users is where we're seeing the real shift.
Combined with Claude for Excel, every function across our business not just engineering is adopting Claude as their primary working tool.