Project summary

I lead a software company devoted to instantiating foundational epistemic infrastructure. We are incredibly long on some of the simplest primitives of cognitive work: agents comparing entities across high-dimensional subjective qualities so as to flush out agent-interoperable, algebraically composable numbers, and caching judgements in fast relational databases for reuse. We will be shipping concrete APIs that support not just extremely powerful search, but timeless knowledge emission by agents.

Recently we released alignment-explorer.exopriors.com. It's a SOTA search tool over a continuously growing alignment-adjacent corpus (e.g. continuously ingesting all LessWrong, EA Forum, community-archive.org, soon all Arxiv articles and HackerNews content). It's a readonly-SQL + compositional vector search API that Claude Code w/ Opus 4.5 is exceptionally good at using. The easy compositional vibe search (vector mixing) functionality combined with long query time limits is remarkably powerful.

What are this project's goals? How will you achieve them?

ExoPriors builds and maintains the DNS of subjectivity. We will achieve this because we are devoted to feeling out and respecting the simplest primitives of cognitive work: agents comparing entities across high-dimensional subjective qualities so as to flush out agent-interoperable, algebraically composable numbers, and caching judgements in fast relational databases for reuse.

Out of several concrete APIs and services we have in motion, one natural consequence soon will be supporting: researchers around the world easily making private rating groups/parties where they spawn agents devoted to tracking bio risk threats that do high-quality, nuanced rating of any internet and private documents, in a way that genuinely and unambiguously factorizes cognitive work in an incrementally useful, non-slop way, with agents spawning when there is a new entity in their top-k ranking for a gated geometric combination of interpretable attributes.

How will this funding be used?

Claude Max subscriptions, API credits for foundational experiments and bootstrapping value, basic infra costs, and a modest founder salary to support freedom of movement and somatic regulation.

Who is on your team? What's your track record on similar projects?

Just me. I shipped alignment-explorer.exopriors.com in a week, a SOTA search tool over a continuously growing alignment-adjacent corpus, supporting arbitrary SQL and compositional vector search, optimized for Opus 4.5 co-piliting.

What are the most likely causes and outcomes if this project fails?

There's canonical things that most interstellar civilizations build, like spreadsheets and sqlite. This is one of them. If I fail to see this technology well instantiated and adding tremendous value in the ecology, it's most likely because I ended up dying in a nuclear war.

How much money have you raised in the last 12 months, and from where?

$600, family.