Overview

Palet is an agent-like search engine built on Open Context.

[1] Palet's search bar.

You can try Palet for yourself by following this link. Search queries at the moment are limited to call functionality as Palet is still in early development. Keep in mind that every query you submit will generate real results. Meaning that dozens of live phone calls will be made to real businesses.

At the moment, Palet is limited to a very early version of search that requires phone calling. We built it in 10 days. However, we are working on making it better and shipping other search features by early February 2025. The GitHub repo can be found here.

Palet is an agent-based search system. When you submit a query, it employs an exhaustive search strategy to generate a result. This approach is just like what a personal assistant would do—checking different sources, contacting people or businesses, and digging deeper until it finds reliable information. Once it has what you need, Palet provides you with a clear result. It is built on Open Context, which is an open standard for connecting Palet's models to the external systems where data lives—things like search engines, your calendars, and your email inbox.

Think of Palet as the user-facing application (like Google Search), while Open Context is the knowledge graph or data layer. Palet uses Open Context as a kind of file system mapped to each user’s account. This setup is how Palet stores and manages chat context and memory state at scale. We'll be launching a sandbox of Open Context so developers can test it out and build their own applications on top of it.

[2] The three types of searchable information. Search engines today can only crawl and index public online information.

In the future, we Palet will evolve from search engine to an end-to-end personal assistant. A self-search engine of sorts that's available to users via web UI, phone, or text. In other words, you'll be able to interface with Palet like you would a human personal assistant.