Rillor Corporation
Capability statement
October 2026
Rillor builds agentic AI systems and the datasets they depend on, and supplies, finances and deploys private GPU infrastructure.
Core competencies
- Agentic AI systems. Planners, specialist agents and audit steps over large bodies of text and data, with citations checked against retrieved evidence.
- AI evaluation and benchmarking. Black-box tests of LLM agents for bias, preferences and deception, including agents bidding in simulated markets and auctions.
- Forecasting and analytics. Forecasts stated before the outcome and scored out of sample against baselines and random controls.
- Document and regulation intelligence. Regulations, legal decisions, filings and rulebooks turned into structured records linked to their source text.
- Data engineering and datasets. Collection, normalization and provenance for pricing, legal, public-records and communications data, plus custom datasets built to scope.
- Financial market data. U.S. equity and futures trade and quote history from 2018. On-chain perpetual futures order books rebuilt from snapshots and book diffs, with fills, liquidations and funding.
- Private AI cloud and GPU infrastructure. Dedicated AI environments on Kubernetes, deployed and operated on infrastructure the client controls.
- GPU systems, financing and forward agreements. GPU systems specified, sourced and delivered. Financing arranged with lending and leasing partners. Forward agreements for GPU systems and data-center capacity.
- Compute pricing. The Rillor Compute Index methodology covers published GPU cloud prices and availability for 11 GPU classes, alongside GPU system prices.
Differentiators
- Every measured figure has a source and a date. Records keep their source, capture time and every transformation applied.
- Evidence classes kept separate. Specifications, listings, quotes, indicative prices, reported sales and verified transactions are stored and reported as distinct facts.
- Black-box, replayable evaluation. Agents are tested only through inputs and outputs, with pinned model versions, frozen snapshots and logged runs that can be replayed.
- Systems in production. Grellum, agentic legal research in which every citation must match retrieved evidence. The Rillor platform, with tenant isolation verified by 109 access, admission and network tests, 7 Oct 2026. The market data warehouse, with U.S. equity and futures trade and quote history from 2018.