Workflow and task execution engine
SCRIMED turns healthcare workflows into sandboxed, audited, human-reviewed task plans.
CarePath AI, DocuTwin, TrialCore, Sanar AI, PayerIQ, and future services map to synthetic fixtures, sandbox runtimes, TrustQA, human approvals, result validation, and production deny gates.
Agent Sandbox Runtime
Every agent receives isolated memory, files, workflow tools, and audit logs.
SCRIMED Atlas Intelligence Core v1 is a synthetic pilot and enterprise assessment operating layer. Medical-record, claims, payer, wearable, and telehealth workflows remain gated until tenant approval, BAA readiness, live connector controls, durable audit, and human review are approved.
Prior authorization support
Per-task sandbox with synthetic policy, packet draft files, scoped memory, and review-only tools.
- Memory: session, operational, knowledge
- Tools: policy evidence retrieval, packet draft builder, missing-evidence detector
- Audit: task opened, policy cited, packet drafted, human review requested
Referral intake automation
Referral-specific sandbox with routing rules, missing-information checklist, and escalation policy.
- Memory: session, operational
- Tools: referral parser, routing constraint map, review queue generator
- Audit: referral structured, gap detected, routing recommendation drafted, escalation created
Claims and denial risk review
RCM sandbox with claim metadata fixtures, policy references, and reviewer disposition state.
- Memory: session, operational, knowledge
- Tools: denial-risk detector, appeal outline builder, documentation gap mapper
- Audit: claim reviewed, risk signal surfaced, appeal draft prepared, review disposition recorded
Ambient documentation review
Documentation sandbox with draft-only note files, source trace, and signature-disabled output.
- Memory: session, operational
- Tools: draft note builder, source trace mapper, missing context prompt generator
- Audit: draft generated, source trace attached, review checklist created, signature blocked
Task execution engine
Task plans route through Planner, Router, specialist agents, TrustQA, and approval checkpoints.
Enterprise Workflow Assessment
Planner Agent. classify workflow, map value signal, attach governance requirements, select evidence needs
- Approvals: Executive sponsor review, Governance review
- Blocked: live connector execution, patient outreach, payer submission
Prior Authorization Packet Plan
Prior Authorization Agent. identify payer policy source, map missing evidence, draft review packet, attach reviewer checkpoint
- Approvals: Payer or RCM review, Governance review
- Blocked: coverage guarantee, medical necessity determination, autonomous payer submission
Documentation Review Plan
DocuTwin. parse encounter context, create draft-only outline, attach source trace, require clinician review
- Approvals: Clinical review, Governance review
- Blocked: final note signature, EHR filing, autonomous diagnosis insertion
ACCESS-Aligned Monitoring Plan
Sanar AI. map chronic-care track, identify wearable/telehealth signal, define outcome reporting, attach clinician-guided review
- Approvals: Clinical review, Executive sponsor review, Governance review
- Blocked: reimbursement guarantee, autonomous patient monitoring, clinical outcome claim
Governed execution API contracts
contract-only boundary before executable workflow APIs
Review request, response, audit, and denied-capability contracts.Identity and access readiness
production authentication, tenant isolation, role permissions, patient context, service auth, consent, and break-glass decisions
Review identity and access prerequisites before execution moves beyond deny-by-default.Execution attempt readiness
idempotency, durable attempt state, replay, concurrency, failure quarantine, runtime-safety handoff, privacy boundaries, and regional compliance
Review the attempt lifecycle before any governed execution route can create or replay work.Runtime safety readiness
throttle policy, abuse signals, connector containment, emergency shutdown, Watchtower escalation, overrides, restoration, and safety drills
Review runtime acceptance controls before any governed execution route can accept executable requests.Governed execution implementation readiness
locked POST endpoints before production execution
Review auth, identity, idempotency, persistence, audit, privacy, and connector prerequisites.Denied execution audit boundary
metadata-only evidence headers and never-capture policy
Review denied-attempt audit envelope before durable logging is enabled.Audit persistence readiness
storage, retention, access, encryption, incident response, residency, and alerting decisions
Review durable audit logging prerequisites before execution moves beyond deny-by-default.Workflow execution results
deterministic synthetic output evidence
Review result fixtures before live workflow automation.Workflow promotion review
synthetic-only approval before production automation
Review promotion records and retained blocked actions.CarePath High-Risk Follow-Up Routing
CarePath AI
13 checks passed, 0 checks failedDocuTwin Draft Note Review
DocuTwin
13 checks passed, 0 checks failedTrialCore Eligibility Review Queue
TrialCore
13 checks passed, 0 checks failed