Royalty Driven LLM

Source-grounded AI answers with claim-level evidence, visible creator allocation, and settlement controls that fail closed.

Production-stable software, externally attested deployments. RDLLM 1.0 is a stable library and service boundary. A real operator cannot claim production readiness or execute direct settlement until independently signed deployment and payment evidence verifies.

Try It In Five Minutes

python -m pip install .
rdllm-first-run
1. AskA bundled synthetic prompt runs locally. No API key is needed.
2. InspectThe answer shows sources, claim evidence, usage metrics, and candidate allocation.
3. VerifyHashes and settlement status expose what is proven, held, or still missing.

Follow the guided first run

Runtime Example

The bundled example is intentionally synthetic and labels itself as such. It demonstrates the response shape without implying that a provider was called or money moved.

RDLLM first-run terminal showing an answer, source footer, claim evidence, and allocation rows
First-run output. Runtime examples and screenshots are regenerated from the public commands.

Choose Your Depth

Plain-language explanationELI5, simple, non-technical, and technical versions in 15 languages. Live use casesFifteen copyable runtime workflows with source-footers and screenshots. Integrate the APIProvider-grounded requests, response verification, metrics, and failure behavior. Deploy responsiblyTrust stores, external attestations, operator gates, and settlement controls. Read the mechanismAttribution math, evidence limits, policy flow, and receipt structure. Read the white paperConcept, research evidence, threat model, limitations, and evaluation plan.

Trust Boundary

RDLLM proves observable runtime facts: which sources were supplied, cited, matched, displayed, allocated, escrowed, and signed. It does not infer hidden training-data use from a citation. Post-hoc similarity is review evidence only. Public receipts use Ed25519; bundled HMAC artifacts are labelled fixtures.

Open the discovery manifest or see how RDLLM itself attributes papers, standards, code, and contributors.