Order and proof become one case.
Order O2C-247, delivery proof, customer record, and exposure arrive together.
Superprocess AI process orchestration
Automate business processes that need judgment.
AI, rules, systems, and people. One process.
One case · one decision · one record Order-to-cash exception Run · O2C-247
CaseO2C-247
StateReceived
PolicyCredit threshold
OwnerFinance controller
Evidence1 event
One governed case Order received
Why a process
Superprocess makes AI operational.
Complete business operation
Superprocess coordinates the full operationOne AI task
One step Read this invoice. Answer returned
Useful intelligence—not the complete outcome.
Carried inside the process →
The whole job
- 01 Receive A case enters →
- 02 Interpret Agents build context →
- 03 Check Policy sets the path →
- 04 Decide Authority is explicit →
- 05 Act Systems are updated →
- 06 Record Evidence remains
- 01 Order arrives
- 02 Agents check policy and systems
- 03 A person approves the exception
- 04 Systems update and evidence is saved
Context carried throughout PolicyStateAuthorityExceptionsSystemsEvidence
- Software gathers
- Agents interpret
- Policy routes
- People decide
- Systems act
- Evidence remains
How it works
Design the process. Run it. See every decision.
Superprocess keeps state, policy, authority, and evidence attached from start to finish.
Invoice factoring 4 stages · 19 modeled steps · 3 outcomes
Validate Understand Resolve Fund
Scroll to follow all 19 steps Swipe to follow all 19 steps
Run
invoice-factoring-wf-2847
Running
14:22:01 workflow.start
Completed
run_id: wf-2847 · —
14:22:03 read_invoice_package
Completed
credit reasoning model · 1.8s
14:22:05 validate_amount
Retried
2 retries · network · 4.2s
14:22:09 request_human_approval
Waiting on human
conf 0.62 · $28,400 over $25K policy · 2h 14m
— policy_check
Pending
pending
— workflow.complete
Pending
pending
Event-sourced history Replayable · durable
SP
Supply Planner Agent
agent · S&OP process
Deployed
Provider options
OpenAI
planning reasoning model
also available
Anthropic
Mistral
Private LLM
Tool access
4 on · 1 off
SAP
inventory, capacity, orders
Slack
planner questions and approvals
Excel
scenario workbook and assumptions
Optimize API
scenario optimization service
Email
external recipient access
Prompt · NLP
Natural-language task
converted into bounded agent steps at runtime
prompt
You are the supply planning agent for the North America S&OP cycle.
Use approved demand, inventory, capacity, and assumptions sources to compare feasible plans.
Produce three options with service, margin, and expedite-risk tradeoffs.
Summarize the recommendation and wait for planner approval before any system writeback.
retry 3x · 60s timeout
· planner approval before writeback
Pending approvals
· 4
Assigned to you
INV-2847
⚠ POLICY FLAG
Acme Textiles · $147,200
Amount exceeds standard policy threshold ($100,000). Vendor in good standing;
prior invoices cleared in 2–4 days.
Agent confidence 62%
· Waiting 18m INV-2849 Metro Wholesale · Missing PO reference
71%
INV-2851 Alpha Industries · Sanctions match (partial)
58%
INV-2853 Acme Textiles · Possible duplicate (INV-2811)
64%
Routed via policy rules Full audit trail
Trace
wf-2847-trace-9f1e
4.8s total
workflow.execute
4800ms
activity.extract_documents
1200ms
tool.pdf_parse
400ms
llm.classify · claude-sonnet-4-6 · 1,284 tok · $0.009
800ms
activity.validate · retry · 1
2000ms
activity.request_approval · pending
—
Structured run evidence Tokens · cost · eval
Current client engagement · trade finance
Pilot completed. Production implementation underway.
- 01 · Scope Core revenue process
- 02 · Flow Documents to decision
- 03 · Capability Client team changes agents, rules, scripts, and routing
Controlled autonomy
Give AI more work. Keep authority clear.
Low-risk work runs within policy. Material or uncertain exceptions pause for the named approver, with the case attached.
- 01 Assist Prepare the case
- 02 Recommend Propose the action
- 03 Act within limits Execute approved decisions
Automatic recovery
Routine failures retry automatically. Unrecoverable cases pause with full context for the owner.
Human decision point Recommendation, not authority
Exception · INV-2847 Owner review
Policy threshold exceeded
- Agent
- Approve with condition
- Approver
- Finance Controller
- After approval
- Update systems · save evidence
Where it fits
Start where judgment slows the work.
One execution pattern for work that crosses systems, judgment, exceptions, and human authority.
Platform · Built to last
Models change. The process keeps running.
Policy, state, authority, decisions, and evidence live outside any one model.
- Agents
- Governance
- Durable execution
- Observability
Swappable intelligence Choose the right model for each agent step.
- Input Invoice arrives
- Agents Agent prepares case
- Governance Policy checks
- Governance Owner approves
- System action ERP updates
Durable execution Retries, failures, and long waits resume from the last completed step.
Operating context stays
PolicyCase stateAuthorityDecisionsEvidence
Observability Every step traced
- Input
- Agent call
- Policy result
- Approval
- Writeback
Evidence→Owner approval→Next version
Your first process
Start with one process that matters.
Bring one recurring operation with a clear owner, a measurable outcome, and enough judgment or exceptions to make it worth solving.
- It recurs
- It has a clear owner and outcome
- It contains judgment, exceptions, or coordination
A bounded path to a production decision
- 01 Map
- 02 Connect
- 03 Test
- 04 Measure
- 05 Operate
- 06 Decide
Outcome Evidence for a production go / no-go.
Prove one process.