Waterloo Road Pump Station. An operator types a question in plain English; the assistant answers it from plant records and controlled documents, and shows its evidence. This page follows one question through the assistant, names the piece of software doing each job, and marks what already exists.
Status as observed on the live host, 28 August 2026, mid-morning. Every status below was read from the running containers, not from the plan.
Switching the model on was not the end of the work. Five faults surfaced within the hour, and every one had been hidden by the stand-in that replaced the model. The most important: asked how to lift an interlock, the assistant was handing the answer writer the step list — the one thing it must never reproduce — while withholding the title block it actually needed. All five are fixed, each is pinned by a test, and the exam grew from 67 questions to 75 to cover them.
One thing remains outstanding, and it is the one furthest outside this project’s control: a read-only login to the plant historian. Until it exists every figure on screen is a stand-in, and says so.
Seven stages. The important thing to notice: the assistant decides what kind of question it is before it looks anything up, and that decision fixes both which sources it may use and what it is allowed to say.
A web page with a question box and an answer pane, opened in an ordinary browser on the control-room PC. Nothing to install. It is running on the host and has been driven by hand in a browser.
It cannot yet be reached by its name: ai.yokogawa.tech does not exist as a network record, so today it is opened down an engineer’s secure tunnel. Three names are outstanding — the screen, the assistant behind it, and the data translator.
A small, fast AI model reads the question and labels it as one of four kinds — or says it is unsure. It does no answering; it only labels. This is the most safety-relevant step in the whole assistant, because the label decides everything downstream.
Two rules are built in: if it is not confident, it asks the operator to rephrase rather than guessing; and if two labels are close, it takes the more cautious one — a question that is partly advice is treated as advice.
This step now runs. The model account arrived on 27 August 2026 and all four kinds are labelled correctly with high confidence. The dropdown it replaced is gone.
Switching it on immediately exposed a fault nothing else could have found. The worked examples the model is shown returned only the label, not the confidence figure beside it — so the model copied that shape, every question scored zero confidence, and every question fell below the caution threshold and came back as “unsure”. The stand-in had been supplying its own confidence all along, which hid the fault completely. That is the lesson worth keeping: a stand-in proves the plumbing, never the step it stands in for.
Operators say “Pump 02” or “the wet well”. The plant records call those PU-302 and WW-101. This step converts everyday names into the plant’s official ones by looking them up in a list — no AI, no guessing. When it gets a name wrong, the fix is to add the nickname to the list.
Live on the host with 8 pieces of equipment and 56 measurement points loaded, and exercised by real questions travelling the chain.
“How many high level alarms last week?”
Counts and measurements taken from the plant’s recorded history. If there are no records, the answer says so — it never estimates a figure.
“What does the level fault alarm mean?”
Explains a signal, alarm or piece of equipment from the manuals and alarm records, quoting the document it came from.
“How do I lift the interlock on Pump 02?”
Finds the approved procedure and identifies it — document number, revision, date — and quotes its prerequisites word for word. It never writes the steps itself.
“What discharge rate avoids a spill?”
Shows what has been run before, what happened when, and the documented limits — then hands the decision back to a competent person. It never gives a number as the answer.
“Is the pump alright?”
When the sorting step is not confident, the assistant stops and asks a clarifying question. No sources are searched and no answer is drafted.
The larger AI model is used here and only here: to put the gathered evidence into plain English. It cannot go looking for anything else — it only sees what its lane collected. It supplies wording, not facts.
This step now runs, and the split it was designed around held under test. Three faults appeared the moment a real model wrote the words, and every one was the same mistake: the model had been asked for a fact that should have come from the evidence — which revision of a procedure it was, what the document was called, which document a limit came from. Each time it supplied something empty or invented, and each time the rulebook refused the answer rather than showing it.
The fix in all three was the same, and it was not a better instruction: those facts now come from the retrieved document itself, so the wording step cannot influence them at all. It supplies English and nothing else — which is what it was always meant to do, now enforced rather than asked for.
Every draft is checked against a fixed rulebook before the operator sees it: no invented steps for safety-critical work, no recommended setpoint, nothing beyond the evidence actually found, and every claim carrying its citation. A draft that fails is rewritten once and then refused — it is never shown.
These are ordinary software checks written in code, not instructions given to the AI. That distinction is the point: an AI can be talked out of an instruction, and cannot be talked out of a check.
This is the part the current stand-in arrangement proves best. The placeholder answers are put through exactly the same rulebook as a generated one would be, and a refusal is a correct result rather than a fault. The rulebook is running today.
Beside the answer the operator sees how it was reached: which kind of question it was judged to be, what data was counted and over what period, how many records were found, and which documents were cited with their revision and date. Procedural and advisory answers carry a visible banner stating plainly what the assistant did not do.
Two further banners are on screen today: one saying the figures are stand-ins rather than real plant data, and one saying no AI model was involved in the wording. Neither can be switched off from the screen.
Each question, the kind it was judged to be, the sources used, the rulebook result and how long it took are all recorded, so any answer can be reviewed afterwards and any mistake traced to the step that caused it. Live since early on, and recording the trial questions being put through the chain now.
The assistant has no free access to anything. These four, and nothing else.
Turns “high level alarms last week” into an exact query over the plant’s recorded history. The definitions that make an answer right — what counts as an alarm, what “last week” means, what counts as a fill — are written down in files an engineer can read and check, instead of being invented per question. It also keeps ready-made summaries so common questions never hammer the live historian.
The assistant never writes database code itself. It fills in a request form; Cube does the rest.
It is running on the host and answering the data lane’s questions. It reads stand-in figures, because the real historian is still not available — so no number it produces today means anything about the plant, and every answer built on one says so on the operator’s screen. The definitions it uses are the real ones and will not change when the historian arrives; only the source will.
Searches the controlled documents by meaning rather than keyword, so “lift the interlock” finds the right procedure even when the document says “interlock override”. Two rules are permanently on: withdrawn revisions are never returned, and every result carries its document number, revision, date, page and section — a passage that cannot be cited cannot be used.
Search by meaning now works: the documents were indexed on 27 August 2026, once the model account arrived. What it has to search is still the problem — three fabricated demo documents, written so the document lanes had something to find. They carry deliberately impossible document numbers and made-up figures so that nothing in them can be mistaken for plant content.
Procedures are now handled differently from everything else, and deliberately so. Ranking a procedure’s sections by how closely they match “how do I lift the interlock” returned the step list — the one part that must never be reproduced — and left out the title block. So the assistant now identifies which procedure governs the question first, then returns that document’s heading, purpose and prerequisites in order, and withholds its step sections entirely. The wording step never receives them. The rulebook still refuses instruction language independently; withholding is the first line, not the only one.
Every piece of equipment and every measurement point, with the everyday names operators actually use. Deliberately a plain list rather than something the AI works out, so it behaves the same way every time and can be corrected by editing one row.
Not a source of answers — the record of them. Every question and every step it took is stored so the assistant’s behaviour can be audited and improved from evidence rather than impressions.
Two supply lines feed those tools. Neither is finished.
Procedures, manuals, alarm rationalisation and design documents are read, split into sections, and converted into a searchable form. Three habits matter: a numbered step sequence is never cut in half; the document’s type comes from the folder it was filed in, not from a guess; and a person confirms the document number, revision and date before anything is stored. A wrong revision on a procedure is a safety problem, not a tidiness one.
The whole path has now been run on the host, end to end, with a person confirming each header at a terminal — but on three fabricated demo documents, not real ones. That was deliberate: it exercises the reading, splitting, header confirmation and storage without any real document being handled before the reviewing arrangement is agreed. It also runs without the AI model, which is why it could be proven now.
The station’s control system records levels, pump runs and alarms into the plant historian. The assistant reads that history through Cube — read-only, never writing to it, and never touching the control system itself.
Two of the three arrived in the last week. One remains, and it is the one furthest outside this project’s control.
Both AI steps now run, and the documents are indexed for search by meaning. Switching them on was not the end of the work: five separate faults surfaced within the hour, every one of them invisible while the stand-in was in place, and all five are now fixed and pinned by tests. The parts a safety reviewer most needs to examine can now actually be examined.
One difference from what was asked for. Two model deployments were provided rather than three — there is no small, cheap model, so the labelling step runs on the large one. It works correctly, but it costs roughly ten times the estimate per question, because labelling is by far the most frequent call. A second deployment on the same account fixes it; no new account and no new key.
Until this exists every figure the assistant produces is a stand-in. The data translator is running and its definitions are written down, so the switch-over is a change of source rather than a rebuild — but an engineer must independently confirm its first real numbers before anybody trusts one.
All three names resolve and all four addresses now hand the visitor to the sign-in page first. The operator’s page and the assistant behind it were moved onto a single address, because the assistant’s own name does not resolve from inside the plant network and a control-room browser would have failed on every question.
What is still unproven is the part no check from here can reach: an operator sitting at a control-room PC, signed in, getting an answer end to end. Names resolving is not the same as a person succeeding.
Each step ends in a check that must pass before the next begins — so that when an answer comes out wrong, there is one place to look, not four.
| Step | What it delivers | Status | What it needs |
|---|---|---|---|
| 1 | The store — where documents, equipment names and ready-made summaries live | Running now | Done. Live, checks passed. |
| 2 | The logbook — recording of every question, put in early on purpose so every later experiment is traceable | Running now | Done. Live, and recording the trial questions going through the chain now. |
| 3 | The documents — the real procedures and manuals, read in and searchable | Proven and indexed, on demo documents | Reading, splitting, human header confirmation, storage and search by meaning have all been run on the host — on three fabricated demo documents, indexed 27 August 2026. The header now also captures the document’s title and authorising role, because the answer writer was being asked for facts it had never been shown. What is outstanding is the real documents, and someone to confirm each one’s revision and date. |
| 4 | Access to plant history — a read-only login to the historian | Blocked | The historian server, a read-only login and agreed table names. Still outstanding, and now the only blocker — the model account arrived on 27 August 2026. |
| 5 | Cube — the data translator, with every definition written down | Running on stand-in figures | Running and answering questions. Two of its definitions were found to be wrong and fixed while proving it, which is what proving it is for. Two further questions about them are deliberately left until the real historian settles them. An engineer must independently confirm its first real numbers. |
| 6 | The assistant itself — the sorting step, the four lanes, the answer writer and the rulebook | Running, with both AI steps | All four lanes answer end to end on the host with a real model, through the rulebook. Five faults surfaced within an hour of switching the model on — every one of them hidden by the stand-in, every one now fixed and pinned by a test. The exam set grew from 67 questions to 75 to cover them. |
| 7 | The operator’s screen — the question box, the answer, the working | Reachable by name, not yet used by an operator | The name records arrived on 27 August 2026 and all four addresses hand the visitor to sign-in first. What remains unproven is the only part no check from here can reach: a person at a control-room PC, signed in, getting an answer. |
| 8 | The exam — 75 engineer-checked questions run end to end and scored | Now runnable, not yet run | No longer blocked: the model account exists, so the sorting step and the wording can both be scored. Eight questions were added covering the five faults found on switch-on. Pass mark unchanged: 85% correct overall, 95% on the two dangerous kinds, and zero breaches of the rulebook. |
| 9 | Document management by operators — uploading and withdrawing documents from a screen instead of a terminal | Designed, not built | Designed and deliberately held back until step 8 passes. Its settings are in place but switched off. |
Sign-in, network routing and hosting are handled by platform services already running on the demo host. They are deliberately left off this map — they protect the assistant, but they are not part of how it answers a question.