Legacy process mining tool ↗

Route every invoice the moment it lands — automatically, or straight to the right reviewer.

This is a live pipeline, not a mockup. Paste a real (or messy) invoice email below and it runs through actual extraction, business-rule checks, and routing logic — the same path production traffic would take.

The discovery engine

Strategy narrows to the shop floor — this page is the Micro layer

Four ways of seeing the same operation, arranged as a funnel rather than a checklist. Hoshin Kanri sets the widest aperture — why the business is looking at all. Macro (value stream mapping) narrows that to where to look. This page is Micro — what your own system logs already know, read automatically at full transaction scale. Ground (Gemba) reconciles all three against what's actually happening on the floor.

Strategy

Hoshin Kanri — strategy deployment

Leadership's True North: 1–3 year breakthrough objectives, cascaded through the X-Matrix and catchball. Sets why the business is looking at all.

North Star Compiler — coming soon

Not shown on this page

Macro

Value stream mapping

End-to-end lead time, queues, and cost centres — where to look.

Open the VSM tool →

Not shown on this page

Micro

System log analysis This demo

Your own ERP event log, read by the process miner: real path variants, rework loops, timestamps.

Inputs

CaseID · Activity · Timestamp · Resource — raw event export, no interviews required

Outputs

Process map, bottleneck ranking, throughput stats, and — one layer up — automatic invoice routing

Why it matters

Full transaction-level scale, blind to nothing that touched the system — catches rework interviews forget to mention

Ground

Gemba — where strategy meets reality

The unwritten playbook: workarounds, exceptions, know-how nobody documented — reconciled against Strategy, Macro, and Micro.

Gemba tool — coming soon

Not shown on this page

Logs show when a button was clicked. They don't show the workaround that keeps the line running, and neither tells you if it was worth building at all — that's why the loop runs Strategy → Macro → Micro → Ground, then back again.

Executive summary

We mined 500 real invoice cases end-to-end and found a process that mostly works — but loses time in one predictable place. Fixing that one handoff, and automating the triage step in front of it, is the fastest path to a faster close.

6.9h

Average cycle time today across 500 invoices — but the median is 4.6h, meaning a long tail of slow cases is pulling the average up, not a uniformly slow process.

15.2h

One bottleneck stands out: Tax Re-check → Approval Pending averages more than double the process mean. This single handoff is the highest-leverage fix available.

74%

Already clear an 8-hour SLA with no process change. The remaining 26% cluster in rework loops and cross-border tax review — exactly the cases the AI layer below flags for review.

Discovery (process mining) and intervention (AI extraction + routing) were both built and deployed as working software in this sprint — not slideware. Try the live demo below with your own invoice text.

Live demo

Incoming email

unstructured input

Routing decision

extraction → rules → route

Run an invoice to see the extracted payload,
business-rule trace, and final routing decision.

Process intelligence — 500-case discovery run

Cases analyzed

500 invoices

3,237 events mined from the historical log

Avg. throughput

6.9 hours / invoice

median 4.6h — long tail from rework loops

Within 8h SLA

74.2 %

illustrative SLA threshold for this dataset

Worst bottleneck

15.2 hrs

Tax Re-check → Approval Pending

What you're looking at: your process, mapped automatically from raw system timestamps — no interviews required. Each gold dot is a real invoice moving through the workflow; red paths are where time is actually being lost, not where someone guesses it is.

In one sentence: this is your invoice process as it actually runs, reconstructed straight from system logs — so bottlenecks show up as data, not opinions, and get fixed in weeks instead of surfacing in next quarter's audit.

Animated process map of the invoice workflow with tokens flowing along paths, colored green to red by transition speed

Fast transition

Slow transition (bottleneck)

Standard flow (happy path)

Exception / rework path

Node size = case volume

Token = one invoice in transit

Who's approving fastest. A coaching signal, not a blame list — useful for balancing workload once a bottleneck is identified.

In one sentence: shows whether a slow approval step is a people problem or a workload problem, so coaching lands on the right person.

Bar chart of average case duration by approving resource

How big the tail problem really is. Most invoices clear fast — this is the honest picture of how many don't.

In one sentence: the tail is where SLA breaches and audit findings come from — this shows exactly how big it is.

Histogram of invoice case durations in hours

Is it getting better or worse. The earliest signal of drift — before it shows up in a quarterly report.

In one sentence: catches drift while it's still small, days before it shows up as a KPI miss.

Line chart of daily average case duration over the discovery window