Health Hub · the platform
One platform for every agent.
Health Hub is the platform every Wellytics agent runs on, the same platform that powers genomics and discovery. It is built on Forge, our operating system: define any use case as blocks, chain them into pipelines, and a new agent ships in days.
Any source becomes any output.
The pipeline
Anything in, anything out.
Source
Document
Image, PDF, scan.
Source
Text · Voice
Dictation, transcripts.
Source
API · Weblink
Systems and feeds.
Process
OCR / LLM / VLM / Python
Composable reasoning blocks.
Output
Document · Text · API
Ready to use downstream.
Multiple blocks combine into a single composite workflow, so a discharge summary, a trend line and a guideline check can run as one orchestrated pass.
Fragmented data, unified.
A fragmented source flows through chained reasoning blocks into a structured record in one orchestrated pass.
Health Hub browser extension
We don't integrate. We overlay.
Every hospital runs a different patchwork of EMRs, lab systems and scanned PDFs, so clinical AI that demands deep integration dies in the mess. So we overlay instead. The Health Hub browser extension lays the entire agent team over whatever EMR is already on the screen (legacy, custom or off-the-shelf), reading it in context and writing back in place. Any system a clinician can open in a browser turns AI-native in days, and the same team, policy and audit trail extend across an entire hospital network with nothing to migrate.
Install the Health Hub extension
No new server and no data migration. Deployed with your IT, fully permissioned and auditable.
Point it at your screens
It learns the EMR's layout and fields. No API, no HL7 project required.
Live in the workflow
The same agents, acting in place where clinicians already work.
Forge · the engine room
Model-agnostic by design.
Forge is the operating system Health Hub runs on, model-agnostic and deployable anywhere.
Frontier adapters
Adapters for nearly every frontier LLM and VLM. Swap or route between models without re-engineering the workflow.
Three OCR tiers
Classical OCR for clean print, VLM-based OCR for messy scans, and agentic OCR for documents that need reasoning to read.
On-prem option
Every OCR tier can be deployed on-premise, so sensitive data never has to leave the institution.
Any deployment
Any public cloud, private cloud or fully on-premise, to match how your institution is permitted to operate.
FHIR-native, when you want it
Start with zero integration via the browser overlay; add FHIR R4 write-back whenever you want outputs to slot into the wider health-data fabric.
The principle
AI perceives. The rules engine decides. The clinician confirms.
Perception is probabilistic, so it never acts alone. A deterministic rules layer makes the call, and a clinician confirms before anything becomes part of the record. That separation is what makes the agents safe to deploy in a hospital.