Aethis is an AI infrastructure platform focused on delivering verifiable, rule-based decision systems. It enables users to encode rules, policies, or laws as code, producing decisions that are consistent and accompanied by proofs and audit trails. The system is designed to handle complex criteria, such as legislation with exceptions and alternative routes, by mapping out all possible paths through the rules and guiding users toward viable outcomes based on their specific circumstances.
The platform leverages language models at build time to synthesize candidate rules from source texts like legislation, policy, or contracts. Subject matter experts define tests to ensure correct behavior, and the engine refines the rules until all tests pass. At runtime, a constraint solver evaluates the compiled rules against user-supplied facts, with no language model calls in the request path. This architecture allows Aethis to return the same output for the same input every time, ensuring determinism and traceability. Every decision made by the system includes a proof and a source-cited audit trail, supporting transparency and regulatory compliance.
Aethis provides several developer tools for integration and workflow management. These include a command-line interface (CLI) for authoring rulesets, evaluating decisions, inspecting schemas, and publishing from the terminal; a Python SDK offering both asynchronous and synchronous HTTP client access to the decision API; and agent skills for reusable workflows such as authoring, publishing, tracing, and regression checks. The platform also supports integration with systems or agents compatible with MCP, including Claude, Cursor, and Windsurf.
The service is positioned for developers and organizations that require auditable, formally verified automated decision-making, particularly in regulated domains. An example use case highlighted is consumer credit pre-qualification, where rulesets can encode affordability checks. Aethis is developed by Aethis Limited, based in London, England.
In the Inference & model serving space, Aethis takes a focused approach. It focuses on ensuring automated decisions are correct, explainable, and compliant with rules or regulations. It is built as a B2B product for developers building AI-powered decision systems. It runs on the command line and API.
Aethis first shipped in 2026. Development happens publicly on GitHub with 23 commits in the last 90 days. Key capabilities include rule encoding, proof generation, and constraint solver. It exposes integrations via a public API and an MCP server.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do