Get your orders
delivered on time.

Your customer calls before your report does. Patok tells you which commitment is at risk, why, and what to move today.

Every plant absorbs its variability in three ways: time, capacity and inventory. Patok measures all three with your real data and prepares the decision for you.

See how Patok thinks
Start with a file No sensors No ERP changes
Discover more
The problem

You find out about problems once they have already cost money.

The data almost always exists. What is missing is the time to review it every day. By the time someone builds the report, it has already cost you.

Cash stuck — inventory that does not turn, invisible shortages
Margin slipping away — rush buys that cost 8–15% more
A plan that changes daily — resequencing by hand, with no criteria
Meetings without decisions — the spreadsheet gets discussed, not what to do

And what weighs most is not the delay — it is that the plant depends on you remembering everything.

How Patok thinks

A complex plant.
Three levers of control.

Every late order comes down to the same thing: variability arrived and there was nothing to absorb it with.

How Patok thinksTIMECAPACITYINVENTORY

Hundreds of events a day.

Work orders, machines, materials, queues, people, surprises. Each one moves on its own and no board can show them all.

All of it gets absorbed in three ways.

Time, capacity, or inventory. When variability hits, your plant pays with one of the three — even if nobody chose which.

There is no fourth. It is operations physics, not opinion.

When one goes out of balance, an order moves.

A machine loses capacity, the queue grows, and a customer's date slips. Almost always without anyone noticing in time.

Patok tells you which one to move, how much, and today.

Resequencing spends time buffer to give the order back its margin. It is an explicit trade, measured with your real data.

Not a dashboard to interpret: the specific intervention that protects the delivery.

TIME68%CAPACITY74%INVENTORY58%AVAILABLE BUFFERresequenceCOMMITTED DATEON TIME

Representation of the flow model · not a simulation of your plant

The complexity of your plant is not managed event by event. It is managed by deciding how it absorbs variability.

How we work

We start with what you already have. With your technology or with ours.

A spreadsheet, an ERP, a data lake — or nothing yet. The starting point does not matter. What counts is that you leave with decisions.

Excel and spreadsheets Your ERP Emails and reports Your data lake Or nothing yet

The question that always comes up

— Does it integrate with our SAP?

— Your SAP already knows how to email a CSV. That is enough to start on Monday.

No API, no middleware, no IT project, and no vendor onboarding. Any ERP —SAP, Oracle, NetSuite, or one built in-house— knows how to schedule a report by email.

Starting is simple

Three steps, from the first file to a routine in your inbox.

1

You share a file

A spreadsheet, a PDF, or the report your ERP already emails. Through whichever channel puts you at ease.

2

We configure it with your rules

Your operating criteria, your products, your customers. Not a generic model: your plant's.

3

You get the first memo

Findings, risks, and actions. If it helps, we make it a routine — weekly or daily, you decide.

And when the data does not exist?

Any AI can read what you already record.

The trouble starts when the answer depends on something your plant never wrote down.

The question

"Why did the line stop on Tuesday?"

No downtime reasons recorded

There is no possible answer. No AI can read data nobody captured.

The most it can do is tell you production dropped that week — which you already knew.

With capture in place

"Tool change. 1 h 18 min. Third time this month on that same machine."

And with that you can decide: it is not bad luck, it is a pattern.

That is where the platform begins — and only as far as it needs to go to hold up the decision.

What it does every day

Someone should review this every morning.

And nobody does it. Not out of laziness — that is eighteen questions, and the shift starts at seven.

18of 18 reviewed
  1. DeliveriesWhich orders will not arrive on time?12 overdue
  2. FlowWhat is the bottleneck today?
  3. DeliveriesHow much margin is left on each commitment?
  4. FlowHow big is its queue?
  5. DeliveriesHow many did we deliver on time this month?
  6. FlowHow much can I release without saturating it?
  7. InventoryHow many days of coverage do I have left?2 below minimum
  8. MachinesWhere do the non-producing minutes go?1 h 18 unexplained
  9. InventoryHow much do I order this week, and of what?
  10. MachinesHow much time went into setup?
  11. InventoryWhat has not moved in months?
  12. MachinesWhich machine became this month's problem?
  13. QualityWhich defect repeats, and where does it start?
  14. MoneyWhat was delivered and cannot be invoiced?
  15. QualityHow much material went to scrap?
  16. MoneyWhich paperwork is missing, and since when?
  17. QualityWhat was closed without inspection evidence?
  18. MoneyHow much cash is stuck, and at which stage?
Your inbox · 07:00

Today's operation · Aug 4 · 12 overdue, 4 due today

Patok reviews all of them, every day. And it only writes to you when something changed.

Little's LawReorder pointPareto of downtime minutesCoverage by σ√LCONWIPp90 cycle time

You do not need to know any of these formulas. They are needed so the number is right.

And after a few months, the thing that really matters happens: the plant stops depending on you remembering everything.

What you get

The first analysis, with your own data.

Not a demo with someone else's numbers. You send a file your operation already generates and we hand it back to you, read.

  • 3Key findingsWhat your data already says and nobody has had time to read.
  • 3Operational risksWhich commitment is threatened and why, with numbers.
  • 3Recommended actionsWhat to do this week, prioritized and with a suggested owner.
  • +What data you are missingTo hold up that decision continuously, not just once.

First analysis

with your data

Findings

Where your load actually concentrates
Which material triggers your rush buys
How far your real lead time drifts from the promised one

Risks

Which commitments are threatened this month
Which material will run out first
Which workstation is limiting all the others

Actions

What to resequence this weekMR
What to buy before FridayAL
What to start measuring to sustain itJP

+ And what data you are missing to hold up that decision month after month, not just once.

You pick the channel. We start with whatever puts you at ease:

Simple confidentiality agreement Anonymized file Corporate email or secure folder

The cost of waiting does not show up in any report. It shows up in the delivery date.

Send us a file from your operation. We will tell you which of those questions it can already answer.

See how it works

We do this for manufacturing and industrial distribution companies that run their own operation. Scope is defined in a first 30-minute session — this is not open-ended consulting.

The thinking behind the product

Física de Planta

Built on operations physics. Math, not opinion.

Patok does not improvise: it applies the same criteria the engineering team behind Física de Planta teaches and uses in real operations. No Industry 4.0 hype.

PhD in engineeringAirbusRolls-RoyceUNESCOAWSMicrosoft

Background and experience of the engineering team behind the method.

Factory Physics

Lead time, WIP, buffers, and variability with proven theory, not rules of thumb.

Analysis in code

Python and statistics on your real data — reproducible and auditable, not a black box.

Proven in operation

Criteria and memos already used in real manufacturing plants, not in a lab.

Nothing to install

Start from your files and emails; no sensors and no changes to your ERP.

Google Cloud
Firebase + GCP
Real time
Sub-second streaming
The infrastructure

None of the above works with data that does not exist.

That is why Patok also installs the capture, the model, and the operational memory that hold up that decision — at your plant's pace, not at the pace of a giant project.

See the full platform
Patok Academy

Knowledge should not die when the expert changes shift.

Cognitive continuous improvement needs one shared mental model. Learn flow engineering and how industrial AI stabilizes your shop floor in our free course.

Start Flow Fundamentals

Frequently Asked Questions

No. Patok lives alongside your ERP. The difference is that Patok focuses on the live operational flow: work orders, WIP, decisions, events, conversations, and actions. Your ERP stays your system of record; Patok is your system of intelligence.

Not necessarily. We can start with operational capture, QR, agents, documents, videos, or data you already have. How much instrumentation you need depends on the case. Many customers start with no extra hardware at all.

It is a complete adoption path. You can start with training (Flow Coach + Academy), move on to an assessment, and then implement the platform. The order depends on your maturity and your urgency.

That is what the assessment is for. We identify whether your biggest opportunity is on the shop floor (Gemba), in your data (Cognitive), or in the office (Office). The 30-minute adoption session is free and gives you clarity.

Yes. Patok Cognitive can turn scattered information — spreadsheets, PDFs, videos, emails — into a base you can query with AI, always with permissions, security, and scope defined up front.

It depends on scope. Flow Coach is live in minutes. An assessment takes 1–2 weeks. A pilot is scoped by processes, users, data, and the integration required. What matters is starting with the right case, not the biggest one.

Start with a real decision.

You do not need to instrument anything to start. With what your operation already produces, we analyze one commitment at risk and tell you what to move.

Send us a file from your operation. We will prepare the first memo.

See the platform
30 min · No commitment Immediate clarity

horacio@patok.in