We've raised a $10M seed from General Catalyst, Valor, SV Angel, Conviction Embed, Eric + Karim from Ramp, Vipul from Together, Medal, and More
We're at a crossroads in how humanity interacts with AI. We either continue on a path that puts AI increasingly in charge, or create a path where humans remain in the driver's seat. Whether we stay in control depends on how fast we embed intelligence into our existing workflows. Over the next 5 years, we will all exist below the model's API as they delegate to us, or stay in control by leveling up. Our mission is to make humanity exponentially faster, and enable you to master this new, fast-paced world.
As a result, browser based apps like ChatGPT and Claude feel impersonal. They can't see or hear what you're doing throughout the day. They can't help you with your existing workflows—instead, they force you to adapt to their own. To make things worse, companies behind the browsers aren't going to fix this because they've got their own AI apps that they want to push people into. You're already seeing this today with Gemini and Chrome. And while browsers are not the right fit for tomorrow's world, it is empirical that what comes next is built on the web stack - enabling devs to build on open, cross-platform standards, with a mature tooling ecosystem.
The extent to which LLMs are useful to people depends on how well they understand the limitations. But what if we didn't need to write prompts or understand LLMs at all? What if capture can help interfaces understand us, adapt to us, and teach us what they are capable of? It'd unlock a generation of users like our parents, and be increasingly useful as we move towards agents, with capture as the control system.
Inspired by Andrej Karpathy's vision of an LLM OS, we built HighlightAI: A capture-based interface based on a secure, intermediary layer that sits between your OS outputs and an LLM, paired with a Prompt Library, and App Runtime for Web Apps, allowing you to run prompts and your existing web apps as native, context-aware assistants. Your collection of prompts & apps start to feel like your personal AI - one that deeply gets your world.
With native context, many prompts become as powerful as apps. You no longer need interfaces to summarize a call or explain code. But you're glad to have an interface to collect all your meeting notes
We're looking for exceptional, design-oriented Full Stack & ML engineers - [email protected]