# AI reads supplier safety data sheets and fills your SDS template and chemical data, with a person checking every sheet (UK construction chemicals maker, H.Essers)

> AI extracts safety data sheet data into your template: a UK construction chemicals maker cut SDS work from 97.5 to 10.8 hours; H.Essers handles 30,000 a year.

- Canonical URL: https://aiagentsformanufacturing.com/use-cases/ai-safety-data-sheet-sds-extraction-authoring-chemicals-h-essers/
- Department: Office & management (https://aiagentsformanufacturing.com/departments/office/)
- Evidence level: Verified. Real deployment at a named or precisely described company, with a measured result from a primary source.
- Added: 2026-11-13
- Last updated: 2026-11-13
- Publisher: AI Agents for Manufacturing (https://aiagentsformanufacturing.com)
- Manufacturers named: H.Essers (https://aiagentsformanufacturing.com/companies/#h-essers)
- AI tools deployed: SmartSDS (Lean Learning Collective) (https://aiagentsformanufacturing.com/tools/#smartsds-lean-learning-collective)

## Role

EHS and regulatory compliance (safety data sheets and chemical data). Typical users: EHS manager, quality manager, regulatory affairs specialist, chemical warehouse manager, sales or commercial engineer handling chemical products

## The problem

Every chemical product comes with a safety data sheet (SDS) of 16 mandatory sections, and several teams need data from it: storage rules, personal protective equipment, fire protection, transport, and the company's own SDS when it sells the product under its name. The data is copied by hand, which is slow, error-prone and has to be redone each time a supplier updates its sheet. At a UK construction chemicals manufacturer with 130 products (case described by its consultant, company not named), the quality manager spent about 45 minutes per product copying data from the supplier's PDF into the company's branded template, 97.5 hours for the catalogue; outsourcing was quoted at about GBP 150 per document. H.Essers, a Belgian logistics provider in chemical warehousing, processes more than 30,000 SDSs a year, and several departments had to search each sheet manually for the information they needed.

## How AI helps

A generative AI model reads the supplier's SDS (PDF, sometimes in another language), extracts the data section by section and maps it into a fixed structure: the company's own branded SDS template at the construction chemicals maker, or the fields that operations, safety and commercial teams need at H.Essers (protective equipment, fire protection, which chemicals can be stored together). At the construction chemicals maker, a trained person checks every output before it is released. Responsibility for the content of an SDS stays with the company placing the product on the market.

## Step by step

1. List who uses SDS data today and which fields each team needs (storage class, PPE, transport, exposure limits, your own SDS template).
2. Gather a test set of 20 to 30 real supplier SDSs, including poor scans and other languages, with the correct answers filled in by your expert.
3. Test the extraction on that set and measure field by field how many are right, wrong or missing.
4. Keep a person checking every sheet, with the fields that changed or that the AI was unsure about highlighted.
5. Connect the output to where the data is used (chemical register, ERP, warehouse rules) and keep the supplier PDF attached as the reference.
6. Set up a review when suppliers send a new SDS version, and keep a log of who approved each sheet.

## Tools

- SmartSDS, a bespoke AI workflow by Lean Learning Collective: used at a UK construction chemicals manufacturer
- Generative AI SDS extraction model (proof of concept): used at H.Essers
- Other tools to evaluate: SDS management software with AI extraction (for example from EHS software suppliers), the document AI services of your cloud provider, your ERP or EHS system's chemical module
- To try it on one sheet: a supplier SDS and any LLM (ChatGPT, Claude, Copilot, Gemini) with the prompt below, keeping confidential formulations out of public tools

## Result

UK construction chemicals manufacturer (Lean Learning Collective, the consultancy that built the workflow): processing time for 130 products down from 97.5 to 10.8 hours (85%), under five minutes per sheet including human review, about GBP 13,500 saved compared with outsourcing on the first run, built in five weeks; the consultant reports near-zero errors, without a published measurement method. H.Essers (company news): proof of concept completed with results at least as accurate as human interpretation of the SDS; the tool also helps commercial engineers screen every SDS before quoting chemical warehousing instead of relying on general assumptions. No time savings published by H.Essers.

As reported by the source; most figures are vendor or self-reported and not independently verified.

## Human review: cautions

- The construction chemicals result is reported by the consultancy that sells the workflow, and the company is not named; H.Essers reports a proof of concept without time figures.
- An SDS is a legal document (REACH, CLP, WHMIS or OSHA HazCom depending on the market): a qualified person must approve every sheet and every extracted hazard value.
- Supplier formulations can be confidential: check where the documents are processed before using a public AI tool.
- Extraction quality drops on poor scans and unusual layouts; test on your own worst documents.

## Try it yourself (prompt)

```text
You are an EHS and regulatory specialist. I want to extract the key data from a supplier safety data sheet for our chemical register.

I will paste: (1) the text of one supplier SDS; (2) the list of fields we need [for example product name, supplier, SDS version and date, hazard classes and H statements, signal word, components with CAS numbers and concentration ranges, exposure limits, required PPE, storage conditions and incompatible materials, fire-fighting measures, UN number and transport class].

Do this:
1. Fill a table with one row per field: the value, the SDS section it comes from, and the exact text it is based on.
2. Mark every field that is missing, unclear or contradictory between sections.
3. List the storage incompatibilities and the PPE in plain language for the shop floor.
4. If I paste a previous version of the same SDS, list what changed.
5. Give a short checklist for the person who must approve the extraction.

Never fill a hazard classification, limit or concentration that is not written in the SDS. Use only the data I paste in; if something is missing, write 'not found' or list it as a question instead of guessing. Do not send, order, post or change anything in any system: produce drafts for my review and approval.
```

## Related use cases

- [Automation and AI check supplier certificates of analysis against your specifications before material is released (Georgia-Pacific, C.H. Guenther, two anonymous manufacturers)](https://aiagentsformanufacturing.com/use-cases/ai-supplier-certificate-of-analysis-check-georgia-pacific/)
- [AI drafts answers to customer questionnaires and RFPs from a library of approved answers and proof documents (Haworth, Ingredion, Cuisine Solutions)](https://aiagentsformanufacturing.com/use-cases/ai-customer-questionnaires-rfp-sustainability-answers-haworth-ingredion/)
- [AI drafts, checks and translates technical documentation and service manuals (Kistler, Bollinger Motors, Hitachi IES)](https://aiagentsformanufacturing.com/use-cases/ai-technical-documentation-manuals-drafting-translation-kistler-bollinger/)
- [AI price engine predicts which price each customer will accept, so price increases stick without losing volume (Hilti, Fabory, a building products manufacturer)](https://aiagentsformanufacturing.com/use-cases/ai-b2b-price-optimization-price-engine-win-probability-hilti-fabory/)

## Sources

- https://www.leanlearningcollective.com/blog/she-was-spending-12-days-on-copy-paste.-ai-did-it-in-hours
- https://www.essers.com/news/article/automating-chemical-safety-data-extraction-thanks-to-ai/

---
Please cite the canonical URL: https://aiagentsformanufacturing.com/use-cases/ai-safety-data-sheet-sds-extraction-authoring-chemicals-h-essers/
