Collective intelligence, grounded in research
We’re building software that helps organizations stop forgetting what they learn and start reasoning with what they know.
People learn and reason together as work happens. Our software carries that learning forward across teams and decisions, helping the organization act on what it knows and stay connected to what it is trying to achieve.
What we are
A collective intelligence, organizational learning lab.
An organization learns through many people working at once. Knowledge grows across teams, decisions and everyday experience. We research how that knowledge can become a shared basis for reasoning, so teams can work through decisions together while staying connected to the outcomes the organization has committed to.
That reasoning starts as decisions take shape. People can question assumptions, surface disagreements and bring relevant knowledge into the work while it can still change the outcome. Decisions and their results then become part of the organization’s memory, informing the next decision across teams and over time.
We build the software foundations for this loop. People and machine agents contribute to a shared context. Language models, constraint solvers, SMT search, closed-form statistics and fitted models bring different kinds of reasoning. Our research defines how those contributions become a decision, with explicit authority, traceable evidence and clear limits on what each method can establish.
The archive
The research library.
Every paper is a technical report: propositions, proofs, benchmark numbers and a named limitations section. Each has a downloadable PDF and a short-form slide summary. Source behind every claim is shared on request.
Start here
Four questions worth thirty seconds.
Short, standalone explainers that bring each argument into focus.
What is JEPA?
A world model that predicts in representation space instead of pixel or token space — and an open research question for us, not a shipped capability.
Read →Why RAG, LoRA and MCP are not the right approach for organizations
Three popular patterns for extending a model — and why each one, taken as the whole architecture, quietly erases the boundary that makes a decision accountable.
Read →Why context is more important than ever
When agents talk to each other directly, the conversation becomes the architecture. When they only read and write a shared context, the context becomes provable.
Read →What is SMT, OR-Tools, and CVC5?
Three tools for three different kinds of certainty: exhaustive search for counterexamples, provable optimization, and the industrial solver underneath both.
Read →The applications built on this research live at Reflective.
Axioms™, Helms™, Converge™ and Organism™ put this research to work in real organizations.
Visit the company site →