Health Cloud

AI pipelines for
Healthcare & Life Sciences

Accelerate prior authorizations, reach every patient with a care gap, and match eligible patients to clinical trials — all within Health Cloud's compliance boundaries.

Authorization_Request__c Record creation

Prior Authorization Automation

The Problem

Prior auth requests require clinical staff to gather patient history, match against payer criteria, draft clinical justification letters, and portal-submit to payers. Each request takes 45–90 minutes. Denial rates run 15–25% due to missing documentation. Delays in auth delay care — patients wait.

Why not DIY?

The payer criteria API changes quarterly. FlowMason's config-driven HTTP callout means updating the endpoint or payload is a JSON change, not a redeploy.

Pipeline Stages

soql_query (patient + clinical history) http_callout (payer criteria API) llm_call (clinical justification draft) DocumentProvider (attach records) ConditionalFlowControl (auto-submit vs. review) dml_write (status + notes) http_callout (payer portal)

FlowMason Components

PipelineRunnerDocumentProviderConditionalFlowControlFMCircuitBreaker

Realistic Outcomes

Metric Before After
Auth prep time 60 min 8 min
Denial rate 20% 9%
Staff capacity Baseline +3× requests per day

Illustrative based on observed patterns. Your results depend on your data and implementation.

Contact (Patient) FMScheduler — nightly batch

Care Gap Outreach

The Problem

Care managers manually identify patients overdue for screenings, preventive care, or chronic condition follow-ups. Lists are stale by the time they're worked. Outreach is generic — the same message regardless of patient history, language preference, or access barrier profile.

Why not DIY?

Running a nightly batch against thousands of patient records, chunking to stay within governor limits, and personalizing each outreach message requires FMScheduler + PipelineRunnerBatch + ForEach — three separate orchestration layers. Each alone is weeks of Apex.

Pipeline Stages

FMScheduler PipelineRunnerBatch (chunk patients) soql_query (clinical history + gaps) llm_call (gap identify) llm_call (personalized outreach draft) ConditionalFlowControl (SMS vs. email vs. call queue) dml_write + http_callout (send)

FlowMason Components

FMSchedulerPipelineRunnerBatchForEachConditionalFlowControl

Realistic Outcomes

Metric Before After
Patients contacted per week 200 2,000
Outreach relevance Generic template Personalized per patient
Care gap closure rate Baseline +28%

Illustrative based on observed patterns. Your results depend on your data and implementation.

Contact (Patient) New trial criteria loaded or new patient record created

Patient Matching for Clinical Trials

The Problem

Trial coordinators manually read clinical notes to find eligible patients. Match rate is 2–5%. A coordinator can screen roughly 50 patients per week against 10–15 trials. Most eligible patients are never identified — trials miss enrollment targets and timelines slip.

Why not DIY?

The TryCatch pattern — gracefully handling incomplete clinical records instead of failing the entire pipeline — is critical here. Clinical data is messy. FlowMason makes graceful degradation a first-class JSON stage.

Pipeline Stages

soql_query (patient record) DocumentProvider (clinical notes extract) ForEach (active trials — parallel) llm_call (inclusion/exclusion check) TryCatch (missing data handler) ConditionalFlowControl (match score routing) dml_write (Trial_Match__c)

FlowMason Components

ForEachDocumentProviderTryCatchConditionalFlowControl

Realistic Outcomes

Metric Before After
Screening throughput 50/week 500+/week
Patient match rate 2–5% 8–15%
Time to first match Days Hours

Illustrative based on observed patterns. Your results depend on your data and implementation.

Org Chat. Phase H

Chat with your Healthcare data.

Beyond the pipeline patterns above, FlowMason ships a chat surface admins drop on any Lightning page. Healthcare users type plain-English questions; FlowMason generates SOQL, validates through the 8-gate sanitiser, runs under FLS, returns rows. Read-only by default.

Try asking:

  • "show me patients overdue on medication review"
  • "count claims denied this week by reason"
  • "which trials match this patient's diagnosis?"

FLS-aware. Allowlist-governed. Permset-gated. No data leaves the org.

Ready to build Healthcare pipelines?

Tell us your use case. We'll show you exactly which pipeline pattern fits.

Talk to us about Healthcare