Captia Technology
Contact

Interactive tool

Physical AI Readiness Index

Physical AI is AI that perceives, reasons and acts on the real world. In a factory it starts by properly capturing data from the equipment that already exists. This index places a plant on the Physical AI Ladder with fourteen questions about observable criteria, from rung 0 Isolated to rung 3 Acting, and returns the dimension holding it back and the concrete next step.

The starting point is the one that orders the whole series: physical AI does not start at the robot, it starts at the data of the machine you already have. The full framework, with sources and dates, is in the industrial physical AI guide, and the technical layer that holds up the first three rungs is the industrial data platform.

The questionnaire

0 of 14 questions answered

Data capture

1. Where is last Tuesday’s production data?
2. How are machine state changes recorded: running, stopped, fault and waiting?
3. Which clock stamps the data?

Context and model

4. Is there a signal list with asset, unit and rate, and someone who maintains it?
5. What travels with each captured value?
6. How much comparable history is kept?

Real time visibility

7. How is the OEE of a line obtained?
8. What share of downtime is left without a classified reason?
9. What can energy consumption be attributed to?

Decision and action

10. Is there any decision the system proposes or executes from the data?
11. Is the effect of that action measured?
12. What does the system do when data is missing or arrives late?

Governance and safety

13. How is a rule, a threshold or a model changed?
14. How are permissions over equipment and networks distributed?

Answer the fourteen questions to place the plant on the ladder.

The five rungs

The ladder orders what a plant can do with its data. Rung 0 Isolated: data lives in each machine, on a screen or on paper, with no comparable history. Rung 1 Connected: equipment speaks a common language and data reaches one place with time and context. Rung 2 Aware: the plant knows what is happening, with OEE, energy, downtime and quality in real time per line. Rung 3 Acting: the system decides and executes under supervision, and the effect of the action is measured.

Rung 4, Physical, means perceiving, reasoning and acting autonomously on the real world. It is the state of the art in the sector and this index does not certify it: the criteria that would have to be demonstrable at that rung, such as policy evaluation on unseen cases or a safety envelope independent of the learned model, are not observable in a one day visit.

The requirements are cumulative and cannot be skipped: a rung counts as reached when its own criteria and those of the rungs below are met. That is why the most expensive failure mode is trying to buy rung 4 on top of a plant sitting at rung 2. The demonstration works with the demonstration part and the demonstration lighting, and the deployment degrades under real variability, with no action and effect log that would allow it to be corrected.

How the rung is computed

The fourteen questions are spread over five dimensions: data capture (where the history lives, whether machine state is recorded on change and which clock stamps it), context and model (signal dictionary, what travels with each value, comparable retention), real time visibility (where the OEE comes from, downtime reasons, energy attribution), decision and action (what the system decides, whether the effect is measured, what happens when data is missing) and governance and safety (rule versioning and approval, permission split and network segmentation).

Each option describes a concrete practice and scores 0 to 3. Each dimension reaches the rung matching the whole part of the average of its answers, and the plant sits at the rung of its most lagging dimension. The result is not a score, because an average would hide exactly what decides the outcome: the dimension that blocks the rest. Answers are processed in your browser and are neither sent nor stored.

This index looks at one thing only: whether plant data holds up what you want to put on top of it. For a broader picture of the operation, covering management systems and administrative processes, the framework is a different one and lives in the digital maturity test. The technical terms the questions use are defined in the industrial glossary.

From rung to plan

The index gives a snapshot and a next step; contrasting it requires looking at the real data. The first three rungs rest on the same technical layer: acquisition and contextualisation with Captia Connect, per line visibility with Captia.ai, and a supervised closed loop. To contrast the result against your own signals, tell us about the case.

Frequently asked questions

What is physical AI?
Physical AI is AI that perceives, reasons and acts on the real world. In a factory it starts by properly capturing data from the equipment that already exists. That is the order this index measures: first the data substrate, then visibility, then the closed loop.
What does the Physical AI Readiness Index measure?
It measures fourteen criteria observable during a plant visit, spread over five dimensions: data capture, context and model, real time visibility, decision and action, and governance and safety. It does not ask about plans or intentions, but about practices you can check with a query, a log or a document.
What are the five rungs of the Physical AI Ladder?
Rung 0 Isolated: data lives in each machine, on a screen or on paper, with no comparable history. Rung 1 Connected: equipment speaks a common language and data reaches one place with time and context. Rung 2 Aware: the plant knows what is happening, with OEE, energy, downtime and quality in real time per line. Rung 3 Acting: the system decides and executes under supervision. Rung 4 Physical: perceiving, reasoning and acting autonomously on the real world, which is the state of the art in the sector.
How is the rung computed and why is there no score?
Each answer scores 0 to 3 according to the practice it describes. Each dimension reaches the rung of the whole part of its average, and the plant sits at the rung of its most lagging dimension. An average score would hide exactly what matters: a level counts as reached when its own criteria and those of the levels below are met, so one weak dimension blocks the rest.
Why does the index never place a plant at rung 4?
Because rung 4 is not certified with a questionnaire. It requires quantitative evaluation of the policy on cases absent from training, a safety envelope independent of the learned model, replay of the episode that caused a failure, and a drift metric with a review threshold. This index places a plant between rungs 0 and 3, the ones observable during a visit, and describes rung 4 as the state of the art in the sector.
Do I need to sign up or leave my email to use it?
No. The index runs entirely in the browser: answers are processed client side, with no sign up, no network calls and no persistence. No answer is sent or stored.