Projects
These are not separate products. They are stages of the same idea: a foundation that turns information into intelligence, a reasoning layer on top of it, and verticals that aim it at a market. Two tools have shipped. The foundation is in active development.
Layer 00 · Foundation
The infrastructure layer everything else runs on. Manifest ingests documents, URLs, and raw text, embeds them into a vector store, runs semantic search, and returns sourced outputs with citations. It is built to be a stranger-demoable intelligence engine, not a chatbot.
Try the live demo → Read the architecture →Layers 01–02 · Reasoning & verticals
Direction, not demos. These are referenced to show where the foundation leads. Public demos and links ship only when each is ready.
Turns retrieval into reasoning: multi-source synthesis, inference, and analysis on top of Manifest. Answers "what does this mean?", not just "what exists?" Architecture defined.
Not yet built · no public link
The first vertical: intelligence applied to athletics recruiting. The foundation aimed at how programs discover and evaluate talent. Proof that intelligence systems generate real value in a market.
Private build · no public demo or data yet
A second vertical, use case being scoped. Named and gestured at, waiting on the vertical pattern to prove out first. Built on the same foundation when it is time.
Concept stage · no public link
Groundwork · Shipped
The earlier work. Smaller in scope, fully finished, and open to use right now. Each taught a piece of what the foundation needed.
Classifies help-desk tickets by category and priority. A FastAPI backend with a hand-built static UI. The lesson: a model only matters once it is a product someone can use.
Open the tool →A domain-specific language that turns symbolic input into visuals and audio. A study in parsers, compilers, and symbolic systems, running entirely in the browser.
Open the tool →The architecture behind Manifest, layer by layer.