sale.order, crm.lead, res.partner
FROM OPERATIONAL HISTORY TO A TESTABLE OPPORTUNITY
Find the next workflow
worth automating.
Explore the evidence. Fill the gaps with a conversation.
Put a defensible value on what could change.
01 / PROTOTYPE
A small working loop.
A clear view of its limits.
Loading the local workspace…
Identifying the data source…
THE STARTING POINT
A business, in its records.
This local dataset represents a Community/OCA operation. Every record is synthetic.
ONE OPERATION · CONNECTED EVIDENCE
l.Operational historyRecords + relationships
Customer recordsAccounts and context
Support casesQuestions and resolutions
Field serviceVisits and follow-up
An illustration of the sample domain; discovered relationships appear in the next step.
DATASET MANIFEST
Synthetic operation
A sample can demonstrate the loop. It cannot establish customer demand, production reliability or achieved savings.
Start with the evidence.
Create an exploration, then run a bounded discovery over this sample.
01
Evidence before inference
See the record references and what is still unknown.
02
Answers with attribution
Staff statements and simulated answers stay labeled.
03
Arithmetic you can inspect
Confirm the inputs before calculating a value case.
LOOK FOR THE WORK, NOT THE HYPE
What the records suggest.
These are candidate opportunities. Select one to inspect its evidence and test its economics; an operational pattern alone is not a business case.
The evidence comes first.
Run discovery from Source data to see observations and candidate workflows.
More observed facts
Candidate workflows
Data quality & relationships
MAKE THE UNKNOWNS EXPLICIT
A conversation with context.
Explain how the work happens and where time goes. Proposed values stay unconfirmed until you review them in the input ledger.
Choose a workflow to explore.
Select a discovery opportunity before starting its interview.
SELECTED WORKFLOW
A VALUE CASE, WITH ITS ASSUMPTIONS
What is worth testing?
Ranking combines the recorded operational signals with confirmed economic inputs. It is a decision aid, not a promise of savings.
No confirmed value cases yet.
Select an opportunity and confirm its input ledger, or load the clearly labeled simulated answers below.
Opportunity comparison
Select a row to inspect inputsEXPLORE THE WHOLE LOOP
Prefer a guided sample?
Load clearly labeled synthetic interview answers for all opportunities, then inspect their calculations and ranking. Use a fresh exploration: existing interview or input data is protected and will not be overwritten.
V0.10 CONTROL PLANE · GRAPH · PODS · DISTILLATION
Lode v0.10 Visual Process Studio, Multi-Agent Architect & Learning Console
Inspect the two-layer Odoo semantic graph and cascade Decimal ROI, customize the 4 ERP Domain Student Pods under locked safety guards, and monitor local Ollama & LM Studio learning telemetry.
Interactive O2C & P2P Semantic Graph DAG
Click a node to highlight upstream/downstream cascade paths
Selected Cascade Flow: O2C (
sale.order -> stock.picking -> account.move) & P2P (purchase.order -> stock.picking -> account.move). Click any node above to inspect its temporal edges and cascade weights.
Primary + Cascade + Portfolio Decimal ROI
SHARED_SYMPTOM_DEDUPLICATION = 0.80Deterministic Python Decimal calculation across verified ontology edges (model_inferred edges strictly excluded).
- O2C Primary + Cascade Range
- AUD 1,800.00 – 2,400.00 (with downstream delivery & invoice cascade)
- P2P Primary + Cascade Range
- AUD 1,500.00 – 2,100.00 (with receipt & three-way match cascade)
- Shared Symptom Deduplication
- 0.80× retention factor on correlated secondary workstreams
Inspect calculation_steps & ontology weights
PM4Py Object-Centric Process Mining (OC-DFG & OCEL 2.0)
pm4py · ocdfg_verified · synthetic_archetype_demoDeterministic Object-Centric Directly-Follows Graph (OC-DFG) bottleneck rankings, p95 sojourn hours, and rework process variants discovered across sale.order, stock.picking, purchase.order, account.move, and res.partner.
Ranked OC-DFG Bottlenecks (p95 & Mean Sojourn Hours)
Loading PM4Py OC-DFG bottleneck rankings (`p95_hours` / `mean_hours` with `upstream_model -> downstream_model` attribution)…
Top Object-Centric Process Variants & Rework Loops
Loading PM4Py OCEL 2.0 process variants and `rework_loop_detected` telemetry…
4 ERP Domain Student Pods & HITL Authority Gates
Select a pod to customize its task focus & Mode B scopestock.picking, fsm.order, helpdesk.ticket
purchase.order, product.supplierinfo, mrp.production
account.move, account.payment
Agent Task & Policy Editor
2A -> 2B DECOMPOSED RUNNERReady to save policy or simulate decomposed 2A -> 2B proposal + Mode B staging.
Local Loopback Runtimes (Ollama & LM Studio)
Ollama Loopback Runtime
Endpoint: http://127.0.0.1:11434
LM Studio OpenAI v1 Runtime
Endpoint: http://127.0.0.1:1234/v1