Industrial AI: from the workshop to on-site fitting, your data drives performance
Technologies & Partenaires
Who will build your project
No anonymous outsourcing: meet the people who will design, build and ship your solution.
Victor K.
Founder & Lead Developer
Prépa HEC, Paris-Dauphine (Finance & Wealth Management). Former UBS. Reserve NCO in the French Army. Full-Stack Developer specializing in Ruby on Rails — leads JAIKIN's technical strategy and product direction.
Patrick E.
Senior Full-Stack Developer
Epitech graduate with 8+ years of software engineering experience, specialized in Node.js and DevOps. Designed an airline ERP system, e-commerce platforms, and production SaaS applications. Owns back-end architecture and systems integration.
11%
of French GDP driven by industry (INSEE)
70%
of administrative tasks that can be automated
15-30%
targeted workshop productivity gain
100%
of your data hosted in France
Why industrial AI, now?
Margins under pressure, skills shortages, dormant data: three forces that make AI decisive for manufacturers
Industry accounts for around 11% of GDP and 13% of employment in France (INSEE). Yet most industrial SMEs and mid-market firms still run on spreadsheet schedules, customer follow-ups done "when someone remembers", and production data lying dormant in machines, purchase orders and the heads of workshop managers.
Three pressures are converging. First, margin: energy, raw materials and subcontracting are getting more expensive, and every hour of workshop delay or every poorly optimised fitting route is paid for in cash. Then labour: the shortage of technical skills is one of the first obstacles cited by manufacturers (Bpifrance Le Lab) — AI does not replace your tradespeople, it relieves them of low-value tasks so they can focus on the craft. Finally, data: you already generate mountains of information (cycle times, scrap, delays, customer satisfaction) that no one uses for lack of a suitable tool.
Industrial AI is not a robot turning up on your shop floor. It is first and foremost a software layer that captures, structures and gives voice to your data so you can decide faster and more accurately — in the workshop as much as in the field.
The legacy of Toyotism, extended by data
Toyotism taught industry two reflexes: eliminate waste (muda) and pull the flow from real demand rather than producing blindly. AI is its natural extension: where lean relied on human observation and kanban cards, AI continuously observes your data to reveal invisible waste, anticipate shortages and synchronise the workshop with real demand. Same philosophy — pull flow, continuous improvement — but at the speed and scale of data.
Data: your new industrial capital
Before AI comes data. And yours is worth its weight in gold — provided you capture and structure it
A generic AI model knows the world; it does not know your workshop, your routings, your real lead times or your customers. The value is not in the algorithm — it has become a commodity — but in the proprietary dataset that only you own: the history of your production, your fittings, your disputes and your successes.
Our first mission is therefore to build this asset. We instrument your sources (ERP, production management, fitting sheets, emails, sensors, customer reviews), we clean and structure them, then we make them usable by AI. This dataset remains your full and exclusive property: it is what will tomorrow feed your automations, your predictions and the preparation of your parts for robotisation.
1. Capture
Connect your scattered sources — ERP/production management, spreadsheets, emails, fitting sheets, machine sensors, field forms — with no re-keying.
2. Structure
Clean, normalise and link the data (products, sites, customers, real times) to turn it into a reliable, queryable base.
3. Capitalise
A proprietary dataset that belongs to you, ready to feed AI, predictions and, in time, manufacturing automation.
Where AI creates value in industry
Six concrete levers, from customer acquisition to on-site fitting — each built on your data
Customer acquisition
Automatic capture and qualification of enquiries, quotes generated from the ERP, orchestrated sales follow-ups. Your pipeline no longer depends on one person's availability.
✗ Before: scattered enquiries, quotes in 2 days, forgotten follow-ups
✓ After: lead qualified and routed in minutes, pre-filled quote, automatic follow-ups
Administrative automation
Invoices, purchase orders, site reports, declarations: OCR reading + AI extraction, checking and entry into your tools. Admin stops swallowing your afternoons.
✗ Before: manual re-keying, amount errors, late paperwork
✓ After: processing in < 2 min/document, automatic checking, zero double entry
Production tracking
Cycle times, work-order progress, scrap and incidents reported in real time on a dashboard. You watch the workshop breathe instead of discovering it at month-end.
✗ Before: progress known in meetings, drifts spotted too late
✓ After: real-time tracking, alerts on drift, same-day decisions
Workshop scheduling
AI-assisted scheduling: machine load, priorities, lead times and available skills. The schedule adapts to disruptions instead of breaking at the first hitch.
✗ Before: frozen Excel schedule, manual rescheduling, bottlenecks endured
✓ After: optimised scheduling, one-click rescheduling, smoothed load
Fitting & delivery routes
Optimisation of fitting and delivery routes: geography, time windows, intervention duration, team constraints. Fewer miles, more jobs honoured.
✗ Before: routes planned on gut feel, delays, wasted miles
✓ After: optimised routes, -15 to -25% miles, customer slots respected
Delivery & satisfaction
After fitting, AI orchestrates the job-completion loop: Google review request at the right moment, follow-ups, satisfaction measurement and detection of at-risk customers.
✗ Before: rare Google reviews, satisfaction unmeasured, word of mouth left to chance
✓ After: reviews collected continuously, NPS tracked, dissatisfaction handled on the spot
Drive by data: dashboard & metrics
What isn't measured isn't improved. A single dashboard, fed continuously by your data
We bring your critical indicators together on a real-time dashboard, plugged directly into your dataset. No more reports rebuilt by hand on Monday morning: the data arrives on its own, consolidated and reliable, ready for decision-making. Every metric is calculated from your systems — no parallel entry.
Production
Work-order progress, machine utilisation rate, real vs theoretical cycle times, incidents.
Quality
Scrap rate, non-conformities, rework, root causes surfaced automatically.
Margin & lead times
Margin per job/product, on-time delivery, quote-vs-actual gaps in real time.
Satisfaction
Google reviews, NPS, after-sales lead time, recommendation rate per customer and per fitter.
Workshop scheduling & route optimisation
Two optimisation problems that AI solves far better than a spreadsheet
Workshop scheduling
AI continuously arbitrates between machine load, commercial priorities, promised lead times and available skills to produce a realistic schedule — and readjusts it the moment a disruption occurs.
- Load smoothing and bottleneck detection before they cause blockages
- Automatic rescheduling in case of breakdown, material delay or customer emergency
- Account taken of the skills and certifications actually available
- Visibility on achievable lead times before committing to the customer
Fitting & delivery routes
For fitting and installation activities, AI builds routes that respect customer time windows, the real duration of interventions and geography — while minimising miles and empty returns.
- Multi-constraint optimisation: geography, slots, intervention duration, teams
- Reduction in miles and fuel (-15 to -25% observed)
- Fewer postponed jobs, customer slots better kept
- Route sheets and GPS points pushed directly to field teams
Automating delivery: reviews, follow-ups, satisfaction
Job completion is a commercial moment of truth. AI makes sure no reputation opportunity slips through the cracks
Google review at the right moment
As soon as fitting is validated, a personalised review request goes out automatically to the customer — at the point where their satisfaction is highest. Your Google reviews become a steady flow, not a stroke of luck.
Orchestrated follow-ups
Job-completion documents, invoices, document requests: follow-ups are sequenced and sent on their own, with the right tone and the right channel, until resolution.
Measured satisfaction
A short questionnaire measures satisfaction after every intervention. At-risk customers are detected on the spot and escalated to your team before a negative review lands.
Sovereignty: your data stays yours
A strategic industrial asset isn't rented from a foreign cloud. It is owned
Your production data, your routings, your prices and your jobs are a competitive advantage. Handing them to opaque platforms hosted outside Europe means risking feeding the competition for free and losing control of your own expertise. We make the opposite choice: infrastructure hosted in France, a self-hosted open-source orchestrator where the context demands it, and a dataset that remains your full and exclusive property. GDPR and AI Act compliance is built in by design, not bolted on afterwards.
- Hosting in France / EU (OVH, Scaleway), end-to-end encryption
- Self-hosted open-source orchestrator possible: zero dependency, data on your premises
- Proprietary dataset: you keep full ownership and portability
- GDPR & AI Act compliance from the design stage (privacy & compliance by design)
Preparing the augmented factory: datasets, parts, robotics
The dataset you build today is the fuel for tomorrow's automation
Structuring your data is not just about dashboards: it lays the foundations of manufacturing automation and robotics. Once your routings, parameters and quality checks are captured in a clean dataset, they can feed assisted program generation, part preparation and, in time, the control of robotic cells. Robotics without structured data is a costly arm that knows how to do nothing; with your dataset, it becomes a natural extension of your workshop. It is Toyotism continued: from data-pulled flow through to the automated gesture.
Today — The proprietary dataset
We capture and structure your operational data: routings, parameters, times, quality checks, incidents. The basis of everything.
Tomorrow — Preparing the parts
This dataset prepares the automation of part creation: assisted program generation, parameterisation, AI quality control.
The day after — Robotics
With data and processes structured upstream, the integration of robotic cells becomes gradual, controlled and profitable.
Our answer: custom software and a dataset that belong to you
Not one more subscription. An asset built for your workshop and your trade
Generic software packages impose their processes on you; mainstream no-code tools can't cope with industrial specificity. We design custom software that fits your routings, your vocabulary and your field constraints — backed by the proprietary dataset we build with you. In the end, you don't rent a feature: you own a digital asset that gains value with every job produced.
Our 5-step methodology
From data diagnosis to continuous improvement — a proven path, designed for industry
Data & process audit
1-2 weeksMapping of your flows (workshop, fitting, admin, sales), inventory of data sources and identification of high-ROI quick wins.
Deliverable : Prioritised roadmap + data mapping
Dataset & proof of value
2-3 weeksBuilding the first foundation of the proprietary dataset and a prototype on your priority case (production tracking, routes or admin).
Deliverable : Initial dataset + validated prototype
Custom development
3-6 weeksSoftware and automations built around your data: ERP/production-management connections, dashboard, scheduling or route optimisation.
Deliverable : Solution in production
Deployment & field rollout
1-2 weeksGradual go-live, training of workshop and field teams, documentation. We bring the tradespeople on board, not just the screens.
Deliverable : Teams trained and self-sufficient
Continuous improvement
OngoingThe dataset grows richer, the models get sharper, new processes are added. The system becomes more accurate with every job.
Deliverable : Gains that accumulate over time
Concrete results in industry
Cases representative of the industrial SMEs and fitters we support
Joinery manufacturer — 60 staff
Manufacturing & workshop
Challenge : Workshop schedule on Excel, progress known in the weekly meeting, lead-time drifts discovered too late. Margin per job estimated by rule of thumb.
Solution : Production dataset (work orders, real times, scrap) + real-time dashboard + assisted scheduling connected to the production-management system.
+22% productivity
Scheduling cut from 4 hrs to 30 min/week. Lead times met at 95%. Margin per job finally visible in real time.
Fitting & installation — 25 fitters
Fitting & field services
Challenge : Fitting routes planned on gut feel, wasted miles, missed customer slots, almost no Google reviews despite good work.
Solution : Multi-constraint route optimisation + job-completion automation (reviews, follow-ups, satisfaction) wired into the field.
-23% miles
23% fewer miles, customer slots kept, and a jump from 12 to 140+ Google reviews in 8 months.
Multi-site SME — 90 staff
Industry & administration
Challenge : Time-consuming admin (invoices, orders, declarations) split across 3 sites, double entry and errors, no consolidated view.
Solution : OCR + AI automation of admin, consolidation of data from the 3 sites into a single dataset, management dashboard.
-70% admin time
70% of administrative time recovered, zero double entry, group reporting available in real time.
Where does your data maturity stand?
Identify your level to target the right initiatives first
Level 1 — Dormant data
0-15% exploité
Your data exists (ERP, spreadsheets, workshop managers' heads) but is neither linked nor used.
Start by: capturing and centralising a first critical source.
Level 2 — Structured collection
15-40% exploité
You centralise some data but control remains manual, after the fact, on spreadsheets.
Next step: build a clean dataset and a first real-time dashboard.
Level 3 — Data-driven control
40-70% exploité
You manage on real-time dashboards; AI assists scheduling and routes.
Next step: automate admin and delivery, enrich the dataset.
Level 4 — Augmented factory
70-95% exploité
Structured data and processes open the way to manufacturing automation and robotics.
Next step: prepare the part datasets and gradual robotisation.
How much does an industrial AI project cost?
Transparent pricing, starting with a fast-ROI data foundation
Data foundation
1 process + initial dataset
3-4 weeks
We capture a critical source, build the first dataset and a high-ROI dashboard or automation. Ideal for proving value.
- Data & process audit
- First proprietary dataset
- Key dashboard or automation
- Team training + 30 days of support
Data-driven workshop
3-5 processes
6-10 weeks
Production tracking, scheduling or routes, admin automation: a workshop genuinely driven by data.
- Consolidated multi-source dataset
- Custom software + ERP/production-management integrations
- Scheduling or route optimisation
- Admin & delivery automation
- 90 days of support
Augmented factory
Enterprise / multi-site
3-6 months
Complete, multi-site data transformation, preparation of part datasets and a robotics roadmap.
- Global data & AI strategy
- Custom multi-site platform
- Manufacturing dataset preparation
- Robotics roadmap
- Priority support for 12 months
Every project starts with a free 30-minute audit. The data foundation is designed to pay for itself within a few weeks, before scaling up.
Industrial AI across France
Data & AI expertise serving manufacturers and fitters, with a reinforced Grand Est presence
Grand Est & Alsace
Our base. Franco-German expertise for industrial SMEs and fitting trades in the Bas-Rhin and Haut-Rhin.
Strasbourg (67) • Mulhouse (68) • Colmar • Haguenau
Auvergne-Rhône-Alpes
Industry, mechanics, fit-out and installation for the Lyon and Alpine basin.
Lyon (69) • Grenoble (38) • Annecy (74) • Saint-Étienne
Bourgogne-Franche-Comté
Metalworking, plastics processing and industrial subcontracting in the region.
Besançon (25) • Dijon (21) • Belfort • Montbéliard
All of France
Remote and on-site support for manufacturers and fitters across France.
Marseille • Montpellier • Toulouse • Nantes • Lille
They trust us
Projects delivered by our team — JAIKIN × Azinove Group
« WhatsApp requests converted into structured purchase orders in seconds, on a custom staffing platform. »
Regional staffing group
HR — temporary staffing
« PricingOS transformed our pricing strategy. »
Lukas Ellmann
CEO, Ellmann Consulting
« Exclusive technology partner for the RadiologyCheck medical imaging platform. »
RadiologyCheck
Healthcare — medical imaging
Describe your project — quote within 24 hours
Personal reply from an expert, no commitment.
Victor Glesskrumhorn
Founder & AI Consultant — JAIKIN
AI implementation and automation expert for SMBs and mid-market companies. 50+ automation projects delivered across France, Germany and Switzerland.
Frequently asked questions about industrial AI
No. Most of our clients start out with Excel schedules and scattered data — that is precisely the starting point. Our first step is to capture and structure your existing data to turn it into a usable dataset. You don't need a cutting-edge ERP or a connected factory: we start from what you have and make it useful.
No, and that's a common misconception. In industry, expertise is rare and precious. AI relieves your teams of low-value tasks (re-keying, follow-ups, manual rescheduling) so they can focus on the craft: producing, fitting, inspecting. Above all, our clients find that their teams breathe easier and that the labour shortage becomes less critical, not that headcount is cut.
A proprietary dataset is the structured whole of your operational data: production history, real times, scrap, jobs, disputes, satisfaction. AI models have become commodities; the real value is in this data that only you own. It is this dataset that personalises AI for your workshop, that feeds your predictions and that will tomorrow prepare manufacturing automation. It remains your full and exclusive property.
No, this is a defining choice for us. We host in France/EU (OVH, Scaleway) and, where the context demands it, we deploy a self-hosted open-source orchestrator on your own servers. Your data — which is a competitive advantage — stays under your control. GDPR and AI Act compliance is built in by design.
AI solves a multi-constraint optimisation problem: geographic position of the jobs, customer time windows, real duration of each intervention, team skills and availability. It generates routes that minimise miles and empty returns while respecting customer slots. Our fitter clients typically see 15 to 25% fewer miles and far fewer postponed jobs.
The data foundation (first source captured, initial dataset, first dashboard or automation) is generally delivered in 3-4 weeks and pays for itself soon after. Larger initiatives — full scheduling, multi-site — roll out over 6 to 12 weeks. We always start with a high-ROI case to fund what follows.
Robotics needs structured data to be profitable: routings, parameters, quality checks, tolerances. By building a clean dataset today, you lay the foundations for the automation of part creation (assisted program generation, parameterisation, AI inspection) and then the control of robotic cells. Without structured data, a robot remains a costly, underused investment; with your dataset, its integration becomes gradual and controlled.
Generic software packages impose their processes and quickly hit their limits on industrial specificity (your routings, your vocabulary, your fitting constraints). Mainstream no-code tools can't take the operational load. Custom software fits your reality and, backed by your proprietary dataset, becomes an asset that belongs to you and gains value with every job — instead of one more subscription.