Inside JrvsOS: An Operating System for the Plant

Inside JrvsOS: An Operating System for the Plant

Inside JrvsOS: An Operating System for the Plant

Rodrigo Coronado, Partner and Head of Innovation at Jarvis Consulting Group

Every plant already has the data to explain where its margin goes. It lives in the historian, the MES, the lab, SAP and Maximo, and each department reads its own slice. We asked Rodrigo Coronado, Partner and Head of Innovation at Jarvis, what changes when the whole plant runs on one operating system.


What is JrvsOS Manufacturing?

JrvsOS is how Jarvis applies AI to a company’s own data. It connects the systems you already run into one governed foundation, without replacing any of them, adds the context of your industry and your company, then the agents and tools that work with it.

Our Supply Chain edition is horizontal: it runs one process, procure to pay, that every company has. JrvsOS Manufacturing is vertical. It runs the plant. The foundation follows ISA-95, the industry’s own model (site, area, line, machine), and every tag, production figure, lab test, work order and logbook entry hangs off the machine it belongs to. Operations, maintenance, quality, process engineering, energy and procurement all work from it.

One plant, one OS: the one-minute opener, on sample data. Watch all ten films, one per team.


Why does a plant need an operating system, not another tool?

Because the losses that matter cross departments. A worn part shows up as slower rates for operations, as breaks for quality, as a recipe drifting for process engineering and as an overdue inspection for maintenance. Each department has a tool for its slice. Nobody sees the whole loss, so nobody owns it.

A tool answers one department’s question. An operating system answers all of them from the same plant model, and every new scenario costs less than the last because the data is already connected.

What does it run, day to day?

We built a sample pulp and paper company, with problems planted in it, to show how it works. These are the scenarios running on it today:

Who Scenario What it shows on sample data, last 90 days
Plant leadership Margin Leakage $8.57M lost against the plant’s own best: about $34.7M a year, 5.8% of margin
Operations OEE & Throughput A paper machine’s OEE from 85.6% to 76.9%; breaks from 1.7 to 2.4 a day
Maintenance Reliability $2.13M lost to unplanned stops; an inspection 42 days overdue
Process engineering RCA Workbench A $533,303 loss traced to the first signal that moved
Process engineering Recipe Conformance Freeness out of recipe for 21 days
Quality Yield & Quality $2.54M of yield loss; $1.01M of downgraded pulp on one machine
Energy Energy $353k above best; utilities from $10.46 to $13.67 a tonne in one month
Procurement Supply & Materials The cheapest chips saved $683k, then 8 of 45 deliveries failed the lab and $229k of loss followed

The figures overlap, so they don’t add up to one total. That overlap is the point: the same loss shows up in every lens.

How do the scenarios connect?

Follow one worn refiner plate. Maintenance raised the plate inspection on 1 September, and it waited for approval. On 7 September the refiner’s plate gap moved: the first signal upstream of the paper machine. On 12 September an operator wrote in the logbook that the refiner’s amps were low. Freeness went out of recipe, starch was raised to compensate, breaks went from 1.7 to 2.4 a day and OEE fell from 85.6% to 76.9%. The corrective work order came on 28 September, 16 days after the warning. The episode cost $533,303.

Every department saw a piece of it. JrvsOS sees all of it, on one timeline, because every scenario sits on the same plant model.

Where does AI come in, and how does the plant know it can trust the results?

AI reads what people write: the operator logbooks, the work orders, the shift notes. In the RCA Workbench it ranks the likely causes, shows the evidence for and against each, says how sure it is, and drafts the A3.

The figures come from fixed rules, and the baseline is each line’s own best demonstrated performance over the last 90 days, not a consultant’s target. Any number traces back to a shift, a lab sample or a work order.

Who should be paying attention?

Plant and mill managers, VPs of Operations and CFOs, and the reliability, quality, process and energy leads who own the fix. It matters most in continuous and high-volume operations: pulp and paper, chemicals, food, metals, building products.

Where do you start?

With the scenario that pays first, on one line. For most plants that’s Margin Leakage, because it puts the loss in dollars against your own best and points at the biggest leak. Then the next scenario goes on the same foundation: root cause, reliability, energy, supply. Nothing commits you to the next step.

See JrvsOS Manufacturing run live
Join us for One Plant, One OS: AI Across Operations, Maintenance, Quality and Energy, a 60-minute live demo on sample data, on Wednesday, December 2 at 1:00 PM MT.
Register now →
Questions? Contact us.

Partagez cet article

Accédez à des informations et des événements exclusifs.

Innover, Responsabiliser,
Agir. Ensemble

Accédez à des perspectives et événements exclusifs.

Abonnez-vous aux mises à jour

© 2026 Jarvis Consulting Group. Tous droits réservés.

Innover, Responsabiliser,
Agir. Ensemble

Accédez à des perspectives et événements exclusifs.

Abonnez-vous aux mises à jour

© 2026 Jarvis Consulting Group. Tous droits réservés.

Innover, Responsabiliser,
Agir. Ensemble

Accédez à des perspectives et événements exclusifs.

Abonnez-vous aux mises à jour

© 2026 Jarvis Consulting Group. Tous droits réservés.

Innover, Responsabiliser,
Agir. Ensemble

Accédez à des perspectives et événements exclusifs.

Abonnez-vous aux mises à jour

© 2026 Jarvis Consulting Group. Tous droits réservés.

Innover, Responsabiliser,
Agir. Ensemble

Accédez à des perspectives et événements exclusifs.

Abonnez-vous aux mises à jour

© 2026 Jarvis Consulting Group. Tous droits réservés.