Turn a result into a plan.
A lab result tells someone what their numbers are. It doesn't tell them what to do about it. Clinical Reasoning interprets the panel and the protocol builders generate a personalized plan to move the markers that are out of range — for the ordering practitioner, or for the consumer who bought the test themselves.
2
products, two buyers — a practitioner tool and a consumer-facing one, on the same engine
Deterministic
mapping from biomarker to functional range to condition to protocol — rules, not model improvisation
Cited
evidence behind every mapping, so the interpretation is inspectable
The result is the product. The plan is what the customer actually wanted.
Both buyers have the same gap at the same moment — a panel comes back, something is out of range, and nothing in the product says what to change.
A result and a reference range
For practitioner-facing labs: the ordering clinician gets the values and does the interpretation and planning themselves, off-platform. The lab's product ends at delivery, and so does its role in the outcome.
For D2C (direct-to-consumer) labs: the customer gets a dashboard of numbers, some flagged red. They search for what to do, act on whatever they find, and often never re-test — there is nothing to retain them.
A result, an interpretation, and a protocol
Practitioner product: Clinical Reasoning delivers findings mapped to possible conditions with citations, and the protocol builders generate a plan the clinician edits and signs — inside your portal, as part of your offering.
Consumer product: the same engine produces a personalized protocol, with a practitioner reviewing it before it reaches the customer, and a natural reason to re-test in 12 weeks to see the markers move.
The modules this uses.
Interpretation plus protocol generation — and a patient surface where the product is consumer-facing.
Clinical Reasoning
Maps biomarkers and patient context to possible conditions deterministically. The LLM (large language model) summarizes findings; the rules engine produces them.
Nutrition Protocol Builder
Biomarker-driven meal plans, foods used as medicine, eliminations, and macronutrient targets personalized to the patient’s diagnostic and goal context.
Supplement Stack Protocol Builder
Evidence-cited supplement stacks across cardiovascular, metabolic, hormonal, inflammatory, and longevity domains. Brand-aware.
Tera Patient Copilot
AI patient portal that coaches patients daily between visits. White-labeled (hosted by Tera) or embedded as an SDK (software development kit) in your patient app.
All ten modules are available — these are the ones this use case starts with.
How it works.
The same pipeline serves both products; what differs is who reviews the output and who receives it.
Result lands
A completed panel passes into Clinical Reasoning from your LIS (laboratory information system) or result pipeline.
Findings generated
Values map to functional ranges and possible conditions deterministically, each with a citation.
Protocol built
The builders generate a plan targeting the out-of-range markers — nutrition, supplements, and the other domains as indicated.
Practitioner signs, patient acts
A clinician approves before anything reaches the customer. The plan runs daily, and the re-test shows whether it worked.
Safe AI you can embed.
Clinical defensibility is the precondition for embedding AI into any healthcare platform — three architectural pillars built to pass your compliance, legal, and clinical leadership reviews.
Practitioner-supervised by design
Tera AI generates; the licensed clinician approves. Every condition hypothesis, every protocol, every adjustment is reviewed and signed off before it reaches the patient.
Reasoning is rule-anchored, not improvised
Tera does not ask a language model to interpret biomarkers. A proprietary clinical knowledge base maps results and wearable metrics to functional ranges deterministically, then to conditions and protocols. The model only summarizes what the rules produced.
Every recommendation is evidence-cited
Three knowledge bases — biomarker to condition, condition to protocol, food to condition and health goal — with every mapping backed by peer-reviewed citations. Audit-ready and inspectable.
Three ways to deploy.
Pick the integration depth that matches your engineering capacity and brand requirements. All three deliver the same clinical AI underneath.
— I — API ONLY
Server-to-server REST API
You build the UI (user interface) inside your own application. Authentication via OAuth or API key. Modules return structured JSON your app renders natively.
— II — EMBEDDED SDK
Drop-in React components
Your brand, your styling, our clinical logic. Embed the modules as React components in your existing application without building the interface from scratch.
— III — WHITE-LABEL HOSTED
We host the interface; you brand it
Tera hosts the full practitioner portal and Patient Copilot. Custom domain, custom theme, our infrastructure — no engineering work on your side.