Implemented process inventory
Executable operating procedures, not static flowcharts.
The process catalog translates domain work into governed orchestration. Each process has a purpose, trigger, inputs, tasks, decision points, owners, controls, outputs, business outcome, and evidence record.
Core process capabilities
Coordinate systems, people, agents, evidence, and authority in one runtime.
Start work from what actually happens.
Processes can begin from schedules, data changes, external events, messages, thresholds, user actions, or governance conditions.
- Event-driven initiation
- Scheduled and recurring execution
- API and message triggers
- Metric and ruleset triggers
Keep accountable people inside the workflow.
Assign review, approval, investigation, decision, signature, clarification, and exception tasks with role, due date, evidence, and escalation.
- Role-based work queues
- Approval and sign-off
- Reason and evidence capture
- Delegation and reassignment
Execute bounded work through governed agents.
Lead and Worker Agents can retrieve, analyze, draft, monitor, reconcile, and act within explicit task contracts.
- Skill and tool assignment
- Input and output contracts
- Authority ceilings
- Sentry preflight and release control
Separate deterministic logic from judgment.
Use rulesets and decision services for deterministic outcomes, while routing ambiguous or consequential decisions to qualified humans.
- Ruleset evaluation
- Decision tables
- Branching and conditions
- Human adjudication
Manage deadlines, waiting states, and escalation.
Timers, reminders, due dates, service levels, aging thresholds, and escalation paths remain part of the process definition.
- Boundary timers
- SLA monitoring
- Escalation chains
- Pause, resume, and expiry
Make process outputs reconstructable.
Every form, analysis, document, decision, approval, exception, signature, and downstream action remains linked through Evidence Data Trace.
- Artifact versioning
- Source and decision lineage
- Audit evidence capture
- Retention and reuse
Lifecycle process domains
One orchestration engine across regulated operations.
Research and translational processes
Target review, evidence synthesis, assay decisions, preclinical risk assessment, candidate selection, and IND-enabling readiness.
Trial design and operations
Protocol review, site selection, enrollment monitoring, deviation management, data review, analysis readiness, and study closeout.
Case, signal, submission, and commitment workflows
Safety intake, signal management, aggregate review, regulatory response, submission readiness, and authority commitments.
Deviation, CAPA, change, and release
Investigation, root cause, corrective action, change impact, method validation, batch disposition, and effectiveness review.
RWE, HEOR, and payer processes
Evidence question intake, study planning, model review, dossier readiness, payer response, and access implementation.
Insight, information, publication, and follow-up
Medical insight capture, scientific response, publication planning, education, commitment follow-up, and lifecycle adaptation.
Process execution lifecycle
Every process follows a controlled, observable sequence.
STEP 01
Trigger and instantiate
Create the process instance with context, scope, product, market, priority, risk, and ownership.
STEP 02
Sentry preflight
Check permissions, evidence, controls, applicable policies, required roles, and execution eligibility.
STEP 03
Assign and execute
Route tasks to AI agents, humans, systems, decision services, and scientific engines.
STEP 04
Monitor and escalate
Track progress, SLAs, aging, exceptions, dependencies, retries, and remediation.
STEP 05
Review and authorize
Apply quality gates, functional review, signatures, approvals, or expert adjudication.
STEP 06
Close and learn
Record outputs, decisions, residual risk, commitments, outcomes, and reusable process knowledge.
Process control center
Execution state remains visible from trigger through closure.
StromaOS maintains the current process state, active tasks, evidence readiness, SLA risk, open exceptions, required approvals, downstream dependencies, and outcome status in one governed record.
This allows teams to intervene before deadlines are missed or quality issues become downstream failures.
Model a critical workflowThe Builder
Design processes, agents, reports and forms.
One builder - in natural language, on a visual canvas, or as executable DSL - so the person who knows the work is the person who can change it. Business and scientific teams define and adapt without losing engineering rigor, and every version they produce carries its own approval gate and evidence record.
Describe what the work should do.
Describe a process, an agent’s remit, a report or a form in plain language and StromaOS converts intent, roles, conditions, controls, tasks and outcomes into a governed draft.
Model flow and responsibility.
Drag steps onto the canvas and connect them - triggers, tasks, gateways, timers, human reviews, quality gates, sub-processes and integrations - with the responsibility for each step visible as you build.
Review the definition underneath.
Definition languages give process, agent and report logic a versioned, testable, deployable and inspectable text form, for the changes that are faster to write than to draw.
Reuse qualified operating patterns.
Promote approved definitions into governed templates that can be adapted by product, market, study and risk profile.
Governance inside the Fleet
The Sentry Agent enforces process controls at runtime.
The Sentry Agent determines which controls apply, whether a process may begin, whether a task may execute, whether an exception requires escalation, and whether the final output may be released.
Validate eligibility before execution.
Check identity, evidence, source systems, required roles, permissions, risk profile, and process version.
Monitor work as it happens.
Evaluate metrics, rulesets, timers, quality checks, dependencies, retries, and policy conditions continuously.
Route deviations into governed resolution.
Pause, retry, remediate, escalate, reassign, reject, or invoke human authority with full context.
Prevent incomplete or unauthorized closure.
Require evidence, approvals, signatures, quality gates, and downstream obligations before completion.
Platform outcomes
Processes become faster because work, evidence, controls, and authority stay connected.
Run the same governed process across teams and markets.
Reduce local variation while preserving approved adaptations and authority.
See where regulated work is waiting.
Track active tasks, aging, bottlenecks, SLAs, exceptions, and approvals before milestones slip.
Preserve the complete execution record.
Show who acted, what evidence was used, which controls applied, what changed, and who approved.
Automate more work without removing human accountability.
AI agents execute bounded tasks while the Sentry Agent enforces controls and humans retain authority.
Start with one regulated process that is still coordinated through email, spreadsheets, and meetings.
Define the trigger, tasks, roles, AI agents, evidence, controls, SLAs, quality gates, authority, and outcome. Then deploy it as a governed executable process.