Industrial IP
drawings, processes, know-how
RequiresIndustrial IP is the core asset; it cannot be exposed to an external model.
HelmcodeOn-premise with zero logs: open weights run inside your walls and never train anyone’s model.
industries · Industry & manufacturing
Your industrial know-how, inside your perimeter
Drawings, processes and know-how are industrial IP: they cannot leave for an external API. Open models on-premise, with IT/OT integration and NIS2 in mind.
compliance
Compliance, built into the stack.
Every regulatory demand mapped to a platform capability that ships built in, with nothing to configure.
IT/OT
plant and edge integration
RequiresPlant data often cannot leave the operations network.
HelmcodeGemma 4 fits on a single GPU at the edge; MiniMax M3 reads scanned drawings and documents.
NIS2
Directive (EU) 2022/2555 · by subsector, transposition under way
RequiresDepending on subsector, NIS2 brings cybersecurity and supply-chain obligations.
HelmcodeAn auditable, on-premise-capable provider reduces third-party exposure in your supply chain.
This page is an informational overview, not legal advice. For your obligations and the risk classification of each system, consult qualified legal counsel. AI Act Guide →
where machinery law is heading
Machinery law caught up with AI.
The Machinery Directive has been replaced by a Regulation that becomes mandatory on 20 January 2027, and it was written with software and AI in mind. If a model ever touches a safety function on your line, it stops being an IT question and becomes product law.
01
AI safety functions are inside product law
The Commission describes the Regulation as integrating provisions for machinery whose safety functions are AI-powered, and for the cyber-safety of compliance-relevant software, data and safety control systems. The model in that chain is part of the product being certified.
02
Higher risk, different route
The Regulation reworks the conformity assessment of machinery presenting a higher risk factor. Which is to say the question of what you may declare yourself and what needs somebody else to look, changes, and it changes for exactly the machinery that has learning components in it.
03
The file is yours to write
The declaration of conformity and the instructions can now be digital, but somebody still has to write them, and the technical file is the manufacturer’s work with its notified body. Infrastructure cannot produce it. What it can do is be a component you are able to describe and hold still.
European Commission · Regulation (EU) 2023/1230 on machinery Adopted on 14 June 2023 and applicable on a mandatory basis from 20 January 2027, replacing Directive 2006/42/EC. The points above are as the Commission itself summarises the change on its machinery page; the specific category your product falls into, and the conformity route it takes, are for your regulatory team and your notified body. read the report →
use cases
Your most common use cases.
The cases with the most traction in the sector, each with its own page in detail.
Professional copilots A copilot over plant and technical documentation, the entry case, at the edge. RAG over internal knowledge Grounded answers over manuals, procedures and specifications, private. Document extraction Structured data from scanned drawings, reports and maintenance records. Visual inspection Defects judged against your own acceptance criteria, at the line, without a photo of your process leaving the plant. Code migration Control and MES software moved off an unsupported stack, with the know-how in it staying yours. Document analysis Analysis across manuals, specifications and incident history, which is where a plant’s real documentation lives.
Recommended open models.
A starting point per task type. The full guide maps 80 cases to the open model for each one.
- DeepSeek V4 Flash in Helmcode
- Volume workhorse for documentation and extraction.
- Gemma 4 12B
- Fits on one GPU at the edge, inside the plant.
- MiniMax M3
- Reads scanned drawings and documents directly.
We are distilling and quantizing these open models into small, tightly specialised versions, trained for one task rather than for all of them. A model like that runs on less hardware, answers faster and fits where the big one does not, your own datacenter included. If you have a process with volume and stable criteria, that is the conversation we want to have with you.
// faq
Questions, answered.
What the sector's technical, compliance and business teams ask.
Does our know-how stay confidential?
Yes. On-premise with zero logs: open weights run inside your walls, so drawings, processes and specifications never leave and never train an external model.
Can it run inside the plant / OT network?
Yes. Gemma 4 fits on a single GPU at the edge, so plant data can stay inside the operations network.
Can it read scanned drawings and documents?
Yes. MiniMax M3 is multimodal and reads the image of a drawing or document directly, returning structured data.
Does NIS2 apply to us?
It depends on your subsector; NIS2 (Directive (EU) 2022/2555) is still being transposed in Spain. An auditable, on-premise-capable provider reduces third-party exposure regardless, and the concrete obligations firm up as the transposing law is enacted.
// get started
START BURNING TOKENS
Skip the AI infra work. Deploy your first private inference endpoint today.
Flat rate. EU data. OpenAI API compatible.