Evidence-based development
Start with the real document, the actual workflow, and a result you can reproduce. Let what you learn change the implementation.
Careers
Build knowledge systems and AI agents
We build tools that preserve work knowledge, incorporate corrections, and help people use AI for real tasks.
The work ahead
Document processing, retrieval, and agents in one product
A document parser, an ontology, an agent, and an editor each solve a different part of the problem. Making them work together as a dependable whole is the work that matters.
We care about the source of an answer as much as the answer itself, and the correction a person makes as much as the first result.
How we approach the work
Engineering and product standards
Performance and usability
A strong retrieval system only matters if someone can use it. Consider performance, interaction, language, and recovery together.
Clear decisions and documentation
Make decisions understandable. Record the reasoning, name the tradeoff, and build systems that can be inspected and improved.
Open positions
Current openings
There are no public openings right now. When a role is ready for applications, we will publish its scope and requirements here.