Engineering & AI consulting · Mechanical and plant engineering

You know AI can give your engineering capacity back. The question is: where to start?

We analyse your data and your structures in detail. Together we identify the two or three cases that tie up the most engineering capacity. And we implement the first one so that it becomes the foundation of your AI structure. So your engineering develops again instead of administrating.

Book a 30-minute call How we work
Every answer with a source Inside your software landscape Acceptance against written criteria

Analysed by someone who owned product structures in mechanical and pump engineering himself and knows what separates a good structure from a bad one. A head of development from the line, not a consultant and not a pure software person.

01 · Starting point

Engineering is helpdesk and tool maintenance in one.

In many engineering organisations every technical clarification runs through the same few people. A design engineer spends hours looking for old calculations, sales waits for an answer, the experienced colleague becomes an internal helpdesk. The knowledge is there, in project folders, emails and heads, it is just not retrievable. What is missing is not skill, but free capacity.

02 · Target picture · Relief in two steps

First make knowledge retrievable, then automate the maintenance.

A knowledge layer makes existing know-how directly retrievable, always with a source. The routine runs automatically, the tricky cases stay with your experts. The freed capacity is the raw material for what makes the difference: new products, faster development, stronger differentiation. Engineering develops again instead of administrating.

The hurdle

But how to approach it?

Are the existing processes and data even suitable? Which area first? What does the first step actually deliver? This is where it fails in most companies, not on the technology.

Our way

The way mechanical engineers develop

Requirements, prototype, acceptance, series production. Success criteria in writing, tested with real questions, go or no-go. No slides, no gamble. The first step is an analysis for EUR 1,200.

03 · Two modes of use

AI in the process, or as a one-off clean-up.

Running in the process

The knowledge agent in daily business

Answers queries from design, sales and service, with sources. Runs permanently and is kept current as new data comes in. Example: sizing knowledge on demand.

One-off clean-up

Clearing what was left undone

The roles are clear: we design everything. Concept, workflow, rules and target structure are developed together with your people. The AI agents only handle the execution: bills of material, master data and legacy projects that used to tie up staff for months, following the defined rules, with sources. Your engineer reviews and approves. Months become days.

Standardisation
40 to 60 %
less time, because execution sits with the agents and the engineer only approves
Why this matters
Work that was shelved for lack of capacity becomes feasible again.
Both can be combined: clean up once, then support the process continuously.

This is exactly the foundation that standardisation and engineering excellence rest on. Only once bills of material, master data and variants are clean do the structural topics hold: product structure, variants, standardisation.

04 · How we work

Five stages. From analysis to the foundation of your AI structure.

Your entry point is a structure and data analysis for EUR 1,200 net. Then the priority workshop, then the proof of concept at a fixed price. By the time you decide on the POC you have invested EUR 4,000, not EUR 40,000. You can stop after every stage. What has been built is yours.

01 · Structure & data analysis

First understand where capacity is tied up

In a conversation you describe what eats up capacity. You share one or two data packages, we analyse them in detail.

EUR 1,200 net · a few daysShow details +

How the analysis works

  • A conversation: you tell us where capacity is tied up and what keeps getting shelved.
  • You release one or two data packages: project folders, bills of material, quotations, documentation.
  • We examine data situation, product structure and workflows in detail, with more than ten years of experience in mechanical engineering and development.
  • Result: a first well-founded assessment. Where capacity is lost, whether your data and structures can carry AI, and what should be cleaned up first.

Framework

DurationA few days
Your effortOne conversation plus one or two data packages
ResultFirst well-founded assessment
PriceEUR 1,200 net
02 · Priority workshop

The two or three cases that tie up the most

Building on the assessment we develop the concept with your team: how to reduce the effort and create a workflow that becomes a competitive advantage.

EUR 2,800 net · 1 dayShow details +

How the workshop works

  • Together with your team: the concept for how the effort is best reduced.
  • Select the 2 to 3 cases that tie up the most engineering capacity.
  • Per case: who uses the agent, which questions it answers, which data forms the knowledge base.
  • Success criteria in writing. The POC will be measured against them.
  • If either side sees no potential, the collaboration ends here.

Framework

Duration1 day
ParticipantsTechnical contact plus decision maker
Result2–3 priority cases with written criteria
PriceEUR 2,800 net
03 · Proof of concept

One week, real data

The most important case on your real data, tested with real questions. Your team assesses and refines it. The result holds even on a no-go.

EUR 9,800 fixed price · 1 weekShow details +

How the POC works

  • You release 50 to 150 documents: project folders, calculations, manuals, internal rules.
  • Read only, no training on your data, nothing leaves the house.
  • Tested with 20 to 40 real questions from your daily business. No demo questions.
  • Evaluation against the criteria agreed in the workshop. Your team assesses the prototype, we refine it together.
  • The result holds even if it is a no: you then know why, and what would have to improve first.

Framework

Duration1 week
Your effort4 to 6 hours
ResultPrototype, data quality assessment, implementation concept, cost corridor, recommendation
PriceEUR 9,800 net, fixed price
04 · Implementation as foundation

In your system, as the basis for everything else

If we decide together to implement, we build the foundation of your AI structure: hosting, model choice, permissions, internal champion.

4–8 weeks · per the cost corridor from the POCShow details +

How the implementation works

  • We make the fundamental decisions together: where it is hosted (EU cloud, semi on-premise, fully on-premise), which models are used, who may see what.
  • Productive data base for a defined user group, every answer with source display.
  • Permissions based on your existing structure, simple maintenance process.
  • Building an internal champion: your people learn to maintain and develop the system themselves.
  • All further cases build on this foundation, without rebuilding the basics.

Framework

DurationTypically 4 to 8 weeks
BasisImplementation concept and cost corridor from the POC
PriceBy security level and scope
05 · Expansion & operation

The next case, the next area

Keep it current, refine it, build on the foundation. The structural topics follow once capacity has been freed.

Monthly retainerShow details +

How operation works

  • The agent runs in your system, we keep it current.
  • New documents and insights flow in continuously, answer quality is monitored.
  • The next priority case from the workshop follows on the existing structure.
  • Once capacity has been freed we tackle the structural topics from the analysis: product structure, variants, standardisation. That is the real lever on margin.

Framework

ModelManaged operation
PriceMonthly retainer by scope
OptionHandover to your own operation possible
05 · Implementation examples

The result of such priority cases: five agents from practice.

These are the cases that tied up the most capacity at other companies, and the agents that came out of them. The same agent, five applications. Each works on your data and answers with sources.

Example 01

Knowledge agent for design engineering

Win design time back. No more hours spent searching for old projects, variants and calculations. Senior staff out of the helpdesk role.

Head of developmentCTOMD
Example 02

Knowledge agent for service

Cut downtime. Fault patterns, spare parts and service history at hand, even when the senior technician is not available.

Head of serviceMD
Example 03

Knowledge agent for product management

Variant knowledge available. The right variant with standard and evidence, consistent product information for sales and distribution.

Head of PMCPOMD
live Example 04

Knowledge agent for technical distribution

A faster inside sales team. Data sheets, spare parts and sizing at hand. More quotations and more revenue per head with the same team.

Head of salesInside salesMD
Example 05

Knowledge agent for compliance & documentation

The work nobody wants to do. The agent prepares the groundwork for risk assessment and documentation, the engineer reviews and approves. Reviewing is faster than writing.

Head of documentationCE officerMD
06 · Proof

From real project data, not from slides.

This is what the result of a structure and data analysis looks like in practice. Plant engineering use case: ten design engineers, around 100 projects, six years of history on network drives, in emails and in people's heads. That is how much capacity sits in searching alone.

Search per query
2 to 5 h
technical clarification
Queries per year
~150
quotations and clarifications
Engineering time per year
~450 h
tied up in searching
POC in that case
1 week
validated there on 5 real queries, today we test with 20 to 40

If the knowledge agent intercepts 70 to 80 % of these queries, more than 300 hours of engineering time per year are freed in this example. That is a typical target, not a guaranteed result. This is exactly why the path starts with analysis and priority workshop: that is where we define in writing what the POC is measured against. After one week you know what the implementation delivers, what it costs, or why it is not worth it.

Technical framework

No foreign software world. Your rules.

Adapted to your tools

ERP, PLM, CAD, documentation, M365. No parallel system.

Microsoft native on request

Azure, M365, SharePoint, Teams, your identities.

GDPR compliant

Processing in the EU. EU cloud to fully on-premise.

Auditable

Every action logged. Source, prompt, result.

Where would AI start in your company?

Tell us in 30 minutes where things get stuck. If it fits, the structure and data analysis for EUR 1,200 follows. If either side sees no potential, we part ways at that point.

Book a 30-minute call →