Proof-carrying optimality for finite identification
Separating candidate construction from independently checkable optimality certificates.
Independent AI research
Meet CIEL. An experimental memory system for AI agents working across long conversations, documents, and decisions.
01 Meet CIEL
Research prototypeCIEL is an experimental memory system for AI agents working across long conversations, documents, and decisions.
Our research focuses on retaining relevant context, handling changing information, and connecting answers to supporting evidence.
02 Selected research
All researchFormal guarantees, measured tradeoffs, and honest limitations. Explore our public studies.
Separating candidate construction from independently checkable optimality certificates.
Testing the quality, throughput, and memory tradeoffs of local training on CPUs.
Exploring quantum–classical methods for attention, optimization, and verification.
Every paper includes its scope and present limitations.
Find us on GitHub03 / Build with us
We’re looking for design partners with long-running coding, research, support, or operations workflows. Bring us a real memory problem.
Start a conversation CIEL is a research prototype.