Lyra LIMS is the governed operating system clinical laboratories run on. It replaces the 15+ fractured vendors a lab stitches together (LIMS, provider and patient portals, billing, integrations, and audit tooling) with one platform where governance is structural, evidence is automatic, and responsible AI is built into the core.

At the center sits the Lyra Engine, a governed-AI kernel that enforces policy, records every action to a tamper-evident ledger, and grounds every model call on the lab's own sealed events. Lyra is proprietary and AWS-hosted, and it runs in production at a working lab today.

Every surface in Lyra inherits the same first principles: enforce policy server-side, hash-chain every action into an append-only ledger, redact PHI on every model call, and keep a human accountable for every release.

The Lyra Engine is the governed-AI kernel every portal runs on. It combines server-side policy control, a set of governed registries, and a Learning AI grounded on the ledger.

Roughly 238 governed policies span 20+ compliance frameworks, each with a dry-run mode so a change can be evaluated before it takes effect.

Integration, instrument, and method registries define what Lyra connects to, which analyzers it trusts, and how each assay is run, all versioned and governed.

A Learning AI grounded on the ledger reasons over the lab's real, sealed events, cites the record, and surfaces policy gaps. It never acts outside policy or takes an irreversible action on its own.

Lyra is one platform with five portals. Every stakeholder in the lab works in the same governed system, on the same ledger.

Every action in Lyra is SHA-256 hash-chained into an append-only ledger. The ledger is the source of truth. Every UI view is a projection of it, never an independent source.

The ledger is externally anchored to immutable S3 Object-Lock storage, so a record cannot be silently altered after the fact. When an inspector or auditor asks what happened and when, the answer comes straight from the record, and the lab can prove a result was never modified.

Responsible AI in Lyra is not a bolt-on. The assistant reasons over the lab's real sealed events and cites the record. It never acts outside policy and never takes an irreversible action on its own. It learns from the ledger and surfaces policy gaps for a human to resolve.

Lyra runs the analytical pipeline end to end, with each step recorded to the ledger:

Instrument results come in through the BioTek Gen5 XPT instrument bridge, which produces an immutable evidence manifest for every run.

Release is governed by a four-eyes rule: approval and release require two distinct authenticated actors, aligned to CLIA (42 CFR §493.1281). The system cannot record a release without two operators on the record.

Access is controlled by RBAC, which intersects explicit per-user grants with the user's persona. A user can only do what both their grants and their persona allow.

Lyra speaks the standards a clinical lab already runs on, projected from the same governed ledger:

Lyra is proprietary and AWS-hosted, on a single EC2 instance running Docker Compose, behind an ALB with ACM-managed TLS and nginx.

bcrypt / PBKDF2 password hashing with TOTP multi-factor authentication.

Signed backups every four hours, with an RPO and RTO of four hours or less.

Lyra is built for the frameworks clinical labs are held to: CLIA, CAP, HIPAA, and 21 CFR Part 11. Governed policy, four-eyes release, PHI redaction, and the tamper-evident ledger map directly to those requirements.

One platform. The whole lab. Governed by design.