Introducing Distill

Over the past year, our team has been working on building Distill, which is tackling a fundamental challenge of our working world -- understanding people and companies.

Distill uses AI to comprehensively summarize a person or company by by reading as much as it can and building a composite picture. For individuals, this includes their professional accomplishments, work, investments, and interests; for companies, it tries to understand their mission, products, and financials.

We've grown the team to 10+ people and are looking for a few more, so I thought I'd take a moment to explain how we got here.

The Technology

Last year, when Russ and I explored what LLMs can do for companies, we tried a lot of things and found many problems they couldn't do - stemming from inconsistencies in reasoning and coding ability. However, there was one thing they could do very well: summarize and categorize text. So, we thought: what if we lean on those capabilities?

• generate search queries based on a bio
• determine which search results are relevant
• summarize the text of each search result
• extract relevant facts and structured data
• and many other small tasks, like figuring out a person's real name(s)

We didn't require a leap of LLM ability in order to be successful, we just needed the patience to deal with all of the complexity of understanding humans. I will say that interfacing with these language models is often mysterious -- hence we've invested a lot into debugging tools and observability (perhaps an article for another day).

The Problem

I've been fascinated by the idea of AI assistance for a long time. A missing piece of building a great AI assistant is understanding the people in your life - who you're meeting with and what your relationship is with them. There's no general-purpose people database on the web - there's Facebook, LinkedIn, Instagram, etc - all closed ecosystems with no API that assistants can access. Plus, you have internal systems (email, calendar, instant messages) that contain your personalized context.

When I imagine an AI-assistant-powered future, I can't help but think that these assistants should know who people are & their relationships with you and each other - that's one of my long-term dreams for what we're building at Distill.

Meanwhile, we've built an incredible tool at providing context on people and companies to help foster deeper and faster connection. It's a tool that founders, investors, executive recruiters, and many more are using daily to be better at their jobs.

In the weeks to come, I plan to share more about what we're building and learning. I hope you'll come along for the ride!

If you want to check it out yourself, please join the waitlist! We're adding people all the time.