What happens to your document
How SpecMake works
A technical PDF goes in. What comes back is a structured product record: each value anchored, wherever its text could be located, to the page and the quote it was read from; an audit of what the extraction missed; a compliance and DPP score, run against EU rule sets we encoded by hand from the regulation text, that names the fields your document is missing; and a flag on a value the document contradicts somewhere else. Source files stay in private storage, reachable only through links that expire after an hour. Here is the pipeline, stage by stage.
The pipeline
Five stages. Translation is the one you choose.
- 1
Extract
Tables are read as tables, so a value stays with its row, its column and its unit, and the markers that matter for compliance (a filled circle in an options matrix, an Ex marking) are not lost on the way in. PDF and DOCX are both supported.
- 2
Structure
The document’s industry is identified (coatings, hydraulics, instrumentation, electrical and others) and the content is structured into one canonical record. Each field carries a confidence score, a reason whenever that confidence is low, and, wherever the text could be located, a source anchor (page number plus verbatim quote) so the value can be checked against the original document in one click.
- 3
Audit
A separate AI pass compares the structured output against the source for completeness, flagging missing fields, internal contradictions, and coverage gaps. Findings are sorted by severity. A value the document contradicts elsewhere is flagged as a conflict rather than silently picked.
- 4
Compliance & DPP scoring
The structured output is matched against EU regulation rule sets: CE marking, REACH (Annex XVII), RoHS (2011/65/EU), ATEX (2014/34/EU), IP rating (IEC 60529), and product safety baseline. DPP readiness is scored against EU field catalogs for the categories in scope today, with more added as the regulation rolls out. All scoring runs on the source document, never on a translation, so the same data sheet gets the same checks and the same result whether it is written in German, Italian or any other supported language. The checks do not depend on the document’s language.
- 5
Translate (optional)
For each target language, descriptive values are translated using the document’s industry vocabulary and your glossary (your own terms and your team’s). A grammar check catches agreement and declension errors. Diagrams pass through unchanged, and the confidence scores and source anchors come from the extraction, not from the translation. Choosing zero languages runs stages 1–4 and returns the record in the source language.
What we built around the model
The pipeline calls a frontier AI model at each stage that needs judgment, but the surrounding system is the product: deterministic engineering that grounds, checks, and constrains what the model returns.
EU regulation rule sets
The regulations covered today (CE, REACH, RoHS, ATEX, Machinery, the Construction Products Regulation, VOC, IP, product safety) are encoded by hand from regulation text, not auto-generated and not lifted from a public list. Domain-scoped so RoHS only fires on electrical equipment, ATEX only on hazardous-area products, the Machinery Regulation only on documents that reference machinery legislation, the CPR only on documents that reference construction products legislation or a Declaration of Performance. The rule sets are verified monthly against official EU sources; when one is updated, stored catalogs are re-scored and changed verdicts flagged.
DPP readiness catalogs
Typed field requirements for the EU Digital Product Passport categories that are currently in scope. Versioned so updates to the regulation can be tracked. Industry-to-category mapping is automatic.
The same checks in every language
Compliance and DPP scoring run on the source document, never on a translation. A data sheet gets the same checks and the same result whether it is written in German, Italian, Polish or any other supported language.
Unit normalization
Dimensional analysis across the unit systems industrial specs use: pressure, length, mass, temperature, and others. 350 bar and 35 MPa compare as equivalent. Dimension mismatches (bar vs kg) are flagged distinctly in supplier comparisons.
Source-quote anchoring
Extracted fields carry a source anchor (page number and verbatim quote) wherever the text could be located, so a value can be opened next to the original PDF page with the exact text highlighted. Audit findings are anchored the same way. Works on saved documents too, not only fresh uploads.
Cross-document alignment
Two-document spec diff and N-way supplier comparison align fields across vendors that name and group properties differently. Proposed matches are checked before they are shown: a match with no basis in either document is dropped, and a field the alignment overlooked is put back, so no source field silently disappears.
Glossary & template memory
Per-user and per-team glossary entries persist across uploads. Custom output templates (logo, layout, field selection) are stored per organization.
Private-by-default storage
Source PDFs and extracted diagrams live in private storage with no public access. Every time a file is shown, the link is created for that view and expires after one hour.
The AI we use
SpecMake calls Anthropic Claude: frontier model, strong privacy posture, EU data-handling alignment. Whenever a new model generation ships, we re-check extraction accuracy on our test corpus of industrial documents and change models only when the result is at least as good.
The choice of vendor is intentional. We chose Anthropic for its safety and privacy posture, the no-training-on-API-traffic default, and its quality on technical-document extraction.
Privacy, in plain language
- We never train models on your data. Anthropic does not train on API traffic by default; we have not opted in to any training program and never will.
- Your documents stay private. Source PDFs and extracted diagrams live in private storage. Every link to a file is created for that view and expires after one hour.
- You control retention. Document deletion is immediate. Account deletion is a two-step email-confirmed flow that cascades all your data; products you uploaded to a team stay with the team.
- Hosted in the EU. Database and storage are EU regions. AI processing runs on Anthropic’s API under Standard Contractual Clauses; details in our privacy policy.
Where SpecMake fits
SpecMake is the data preparation layer. The output is portable; integration is direct.
Pair with a DPP platform
JSON-LD output mapped to schema.org Product vocabulary feeds directly into platforms that handle passport hosting, QR generation, and registry submission.
Pair with your PIM
Akeneo sync is built in: preview the field mapping against your own attribute set, approve it, and push. Other systems connect through structured JSON, XLSX, and CSV outputs or the REST API.
Pair with your audit workflow
Per-document audit findings, per-regulation pass/fail breakdowns, and source-anchored field verification, exported or reviewed in-browser before data enters downstream systems.
Try the pipeline on your own document
Upload a spec sheet or technical document. Structured data, audit, and compliance check back in minutes.