The best AI consultant for a manufacturer in the Netherlands or Flanders is the one that can start from your historian, MES and maintenance records on one line, and hand over a system your own IT and OT teams can support. Disclosure: Crux Digits publishes this list and is listed first. We say why, and who the others are better for.
This guide is for the IT manager, CIO, informatiemanager or OT lead at a plant or manufacturing group of 250 to 5,000 people, in machine building, food, chemicals, metal or high-tech. Crux Digits builds AI systems alongside your ICT partner and your system integrator; we do not run your OT network, your servers or a helpdesk. For the Dutch market across all sectors, see our guide to the best AI consultants in the Netherlands; this page covers manufacturing only.
How did we choose the AI consultants on this list?
We kept firms whose own website shows manufacturing work and a presence in the Netherlands or Flanders, checked on 29 October 2026.
1. Manufacturing on their own site. Use cases written in plant terms, such as vision inspection, predictive maintenance, process data or MES.
2. A presence in the Netherlands or Flanders. An office, lab or hub in either region.
3. Published, checkable specifics. Prices, deliverables or named use cases on their own pages. Where a firm publishes no prices, we say so rather than guess.
4. Clear ownership after go-live. Who owns the code, the model and the running system once the project ends.
5. Honest fit. Each entry names the buyer better served by that firm than by us.
Beyond our own pick, this is not a ranking: the entries after ours are in alphabetical order. No firm paid to be listed, there are no affiliate links, and we left out firms whose own site did not show manufacturing work on the day we checked.
Which AI consultant do we pick for manufacturing, and why?
Our pick is Crux Digits: fixed published prices for the first two steps, and the code, model and test set handed over to you.
Disclosure: Crux Digits publishes this list and is listed first. We say why, and who the others are better for. Every reason below is checkable on our own pages.
Fixed prices, excluding VAT: an AI audit at €2,500 in one to two weeks, a Production-ready MVP at €20,000 in four to six weeks, and production from €50,000, per our pricing page. Each step is a separate agreement, and the audit fee is credited in full if the MVP follows within 60 days. At handover you receive the code, model, prompts and test set, with no licence back to us.
Plant work, described in plant terms. Our manufacturing page covers maintenance planned from PLC and SCADA signals, vision inspection at line speed, process optimisation against MES data, and OEE assembled from machine data instead of a manual tally, deployed on the edge or on-premise with alerts wired into the MES, ERP and maintenance flow. Our published predictive maintenance case study describes a system built for a sugar producer; the client name is withheld pending permission.
Next to your ICT partner, not instead of it. Your ICT partner or internal IT keeps servers, network, identity and patching; your OT team and integrator keep the PLCs, SCADA and historian. We deliver the AI system and the evidence an auditor asks for. We work in English and Dutch from the Utrecht region and also serve Flemish plants, see AI consultant in Belgium.
Who should not pick us: a group that wants one supplier for AI plus managed services and round-the-clock operations, an OEM putting AI into the embedded software of the machine it sells, or a plant with an unsolved research question. The entries below are better for each of those.
Which other AI consultants serve manufacturers in the Netherlands and Flanders?
Seven alternatives passed our checks: DataNorth, Flanders Make, Itility, ML6, Sioux Technologies, Sirris and Xebia, listed alphabetically.
DataNorth. An AI consultancy with offices in Groningen and Amsterdam. Its manufacturing page lists automated quality control, supply chain management, administrative work such as work orders and bills of materials, and predictive maintenance. It publishes a price of €2,900 for its AI consultancy for manufacturing and €15,000 for an AI training and workshop, with experts in Dutch, English and German. Best for: a manufacturer that wants a priced advisory session and staff training as the first step, including at German-speaking sites.
Flanders Make. Not a consultancy but a research centre for industrial innovation, pooling more than 1,100 researchers at the five Flemish universities and its own applied research division, per its about page. Member companies work on pre-competitive topics on a shared platform, even with competitors, and it runs testing, validation and training, such as an AI in manufacturing course in Leuven on 19 November 2026. Best for: a Flemish manufacturer whose question is still research, such as flexible robotics or new sensing, rather than a build on known methods.
Itility. An IT solution provider headquartered in Eindhoven, with offices in San Diego, Austin and Iasi, offering business transformation, (Gen) AI, software and data, IT foundation and managed services. Manufacturing is one of its three target industries. Its manufacturing page brings practice from PLM, ERP and MES implementations and shows cases such as an AWS SageMaker model that predicts blade wear-out to prevent die defects. Best for: a high-tech or manufacturing group that wants AI, the data platform and managed services from one supplier, which we do not offer.
ML6. An AI engineering company with offices in Amsterdam, Berlin, Eindhoven, Ghent and Munich, which also sells an enterprise AI platform, Unum. Its industrial page covers vision quality control, process steering connected to PLCs, predictive maintenance and digital twins, and capturing the knowledge of retiring experts. Best for: a multi-site group in the Benelux and Germany that wants a large AI engineering team, or a platform, across several plants.
Sioux Technologies. A high-tech solutions partner headquartered in Eindhoven, founded in 1996, with more than 1,300 engineers in software, mechatronics, electronics, mathware and assembly. Its data science and machine learning work covers predictive maintenance, machine control through anomaly detection and image analysis, integrated into cloud, edge or embedded systems. Best for: an OEM or machine builder putting AI inside the machine it sells, where the model lives in the embedded software.
Sirris. The collective centre of the Belgian technological industry, founded by Agoria in 1949, with eight locations in Belgium and community hubs in Kortrijk, Ghent, Antwerp, Genk, Brussels and Charleroi. It offers technical advice, industrial labs, training and innovation roadmaps, and its site lists shop-floor AI events and articles. Best for: a Belgian technology company that wants independent advice and lab access before it chooses a build supplier.
Xebia. A consulting, engineering and training firm present in 16 countries, with AI strategy, AI and GenAI solution development, AI managed services and upskilling. Its AI in manufacturing explainer lists predictive maintenance, visual inspection, production planning and process optimisation. Best for: an enterprise group running a multi-team programme across cloud, data, software and AI, with managed services after go-live.
Which kind of firm fits which plant problem?
Match the firm to the problem: a known use case on existing data needs a builder, an unsolved technical question needs a research centre.
A known use case on data you already have. Vision inspection on one line, failure prediction on a pump group, OEE loss attribution from MES stop reasons. This is applied AI, and a builder such as Crux Digits, DataNorth or ML6 fits. Fix the first step's scope and baseline.
AI inside a product you sell. A vision module in a sorting machine, anomaly detection in a drive controller. The model ships with every machine and is part of its conformity file, so a high-tech engineering firm such as Sioux Technologies fits better than an AI consultant.
One supplier for platform and operations across sites. A data platform, MES integration and managed services for several plants. An IT solution provider with managed services, such as Itility or Xebia, fits.
A question nobody has answered yet. Flexible robot cells, new sensing, unknown material behaviour. A research centre such as Flanders Make or Sirris fits, often through shared projects.
The expensive mistake is buying for the third problem when you have the first: a multi-plant platform contract to answer one vision question on one line.
What plant data should be ready before the first consultant visit?
Four exports decide the first weeks: historian values, the asset model, MES context and maintenance records, all for the same assets and months.
From the historian, such as AVEVA PI, Ignition or WinCC: recorded values with quality flags and UTC timestamps, plus the compression or deadband settings per tag, because those settings decide which signal survived. From the asset model: which tag belongs to which pump, press or line. From the MES: orders, batches, stops and stop reasons. From SAP PM or Ultimo: maintenance notifications with malfunction start and end. From the LIMS, where quality is the question: sample time and result.
Our historian export guide lists the fields. For loss analysis, OEE per line from MES and historian data shows the build. OEE combines availability, performance and quality, so a consultant who cannot say which of the three a use case moves has not scoped it yet.
A consultant who asks for one week of exported data before quoting a build is doing it right. One who quotes without seeing any is guessing.
How do the Machinery Regulation and the AI Act affect the choice in 2027?
The Machinery Regulation (EU) 2023/1230 applies on a mandatory basis from 20 January 2027, according to the European Commission.
Machinery placed on the EU market before then must comply with the Machinery Directive 2006/42/EC. For AI in a plant the deciding question is whether an output carries a safety function: a camera that stops a robot cell is a different project from a camera that rejects a mislabelled pack.
The Digital Omnibus on AI entered into force on 27 July 2026, and the Commission now dates the high-risk rules from 2 December 2027 for Annex III systems and from 2 August 2028 for AI embedded in Annex I products, with machinery named as an example.
What this means for the choice: ask every consultant how they classify your use case, and who becomes the manufacturer if an AI function changes a machine. A builder should work inside your or your machine builder's conformity process, with the notified body where one is needed, not around it. Our analysis of AI Act high-risk rules in industry covers camera inspection, maintenance prediction and safety functions.
Who owns what between the consultant, the ICT partner, the OT team and the machine builder?
Five parties touch a plant AI system, and the contract should name an owner for each layer before the build starts.
Your OT team and system integrator own the PLCs, SCADA, the historian and the path data takes out of the OT network, secured under the ISA/IEC 62443 series for industrial automation and control systems. Your ICT partner or internal IT owns servers, network, identity, backups and patching. The machine builder owns the CE marking and the safety functions of the machines it supplied.
The AI supplier owns the model, the code, the test set and the documentation until handover, and you own them after. A process or quality engineer in the plant owns the decision the model supports,.
Ask each firm on this list which of those layers it wants to own. Itility and Xebia offer managed services, so they can take more of them. Crux Digits takes only the AI layer and hands it over to the people who already run the rest.
Which questions should an IT manager ask in the first call?
Eight questions separate a manufacturing AI consultant from a general one; ask them before you request a quote.
1. Which historians, MES and maintenance systems have you worked with, and which tags and tables would you ask for first?
2. What does the first step cost, at a fixed price, and what do we keep if we stop after it?
3. Who exactly builds it, and are they the people in this call?
4. Where will the model run, on the edge, on-premise or in our cloud, and what does our ICT partner have to host or patch?
5. How does data leave the OT network, and which part of that path does our OT team control?
6. Is any output a safety function, and if so, who acts as manufacturer under the Machinery Regulation?
7. Which number moves: availability, performance or quality, scrap rate, or unplanned stops, and how is the baseline measured?
8. What is in the handover pack: code, model, test set, retraining instructions and the classification record?
Where does Crux Digits fit?
Crux Digits fits a plant of 250 to 5,000 people with one measurable problem on existing data and an IT team that wants ownership.
We start with the audit, pick one line or asset group, build the MVP on your exported data, and hand over the code, model and test set with documentation your ICT partner can host. We do not run your OT network, host the system or staff a helpdesk. Where you need that after go-live, your existing ICT partner takes it on, or Itility or Xebia above are the more complete choice.
For how we run applied AI and machine learning projects in fixed steps, see applied AI and ML for mid-sized business.
Last updated: 29 October 2026.
A consultant tells you where AI pays off; Crux Digits also builds it. A fixed price per step, one named expert, from Utrecht.
AI consultancy in the Netherlands →
