Smaller minds.
Better harnesses.
Verifiable memory.
Why Meotis is building chain-native intelligence on the Internet Computer, and why this programme is open to researchers and builders.
Join the research programme.
Small language models, agent harnesses, WebAssembly, quantization, provenance, private memory or on-chain AI: if you build in any of these areas, this programme is yours.
Thank you, your application is on its way. We'll get back to you shortly.
Five research tracks
Runtime
Lower instructions per useful operation: SIMD, quantization, ternary models and session-level optimization.
Models
Task-specialized small language models, evaluated under real ICP constraints, not only on GPU benchmarks.
Harnesses
Orchestration, memory routing, tool use, verification and multi-call strategies under hard budgets.
Proof
Every critical agent action gets an explicit budget, an execution receipt and a traceable memory path.
Cognitive Heritage
Local encrypted vaults, ORIGYN SDK certification and gated access to authenticated human memory.
The AI industry is no longer only racing for bigger models. It is racing for more intelligence per unit of compute. Phones force efficiency for privacy and energy; blockchains force it for consensus and verifiability. Both point in the same direction.
Efficiency is the product
On ICP, a call runs under a deterministic budget: more compute cannot be purchased during execution. We ran the same Qwen 2.5 0.5B model, with the same weights and the same mainnet limit, through two different software paths: 10 tokens per call → 29 tokens per call. No new hardware, no larger model. The model is the engine; the harness is the vehicle.
An open research loop
We published the paper, the code, the raw measurements and the reproduction procedure. Within 48 hours, the maintainer of llama_cpp_canister (OnicAI) independently traced the bottleneck, validated it on mainnet and shipped a ~2.8× faster release. Open research became production infrastructure in days.
Measure. Publish. Reproduce. Improve.
Memory as heritage
Agents do not only need a brain: they need memory that can be trusted. We call Cognitive Heritage the evolving body of knowledge, decisions and creations a person accumulates over a lifetime, today fragmented, with uncertain provenance. Private content stays local or encrypted; ICP anchors authenticity, provenance and permissions; the ORIGYN SDK certifies the rights. Human memory becomes sovereign, programmable and durable digital heritage.
