Consolidated 40,000 pages of SDS safety data sheets and equipment manuals into AI.
Rebuilding safety education and emergency response
SDS/manual search time
Average 10 min → average 15 sec
Reduction rate
97% reduction
Documents covered
40,000 pages total
Industry
Chemicals
Target departments
Manufacturing, R&D, Safety and Environment
Implementation period
6-week PoC → phased rollout over 8 months
A digital foundation for safety
Company D handled hazardous and poisonous substances, and its business required more than 40,000 pages of safety-related documents, including SDS, MSDS, equipment SOPs, and emergency response manuals.
However, it took field workers more than 10 minutes on average to find the information they needed, and during emergencies the default approach was often to ask a senior colleague.
By implementing CLAVI Mining as a digital safety foundation for maintaining zero serious incidents, the company enabled instant access to safety information while improving training effectiveness.
Reasons for selection
The biggest reason Company D chose CLAVI Mining instead of another AI solution was safety assurance in an industry where incorrect answers can directly put human lives at risk. Its patented multi-layer verification technology and instant source citation were evaluated as the only practical way to give field workers a clear reason to trust AI.
In addition, on-premises support allowed the company to keep technical information subject to export regulations inside the organization, while multilingual support enabled overseas sites to use the same information foundation. In a prior security assessment meeting, the IT and legal departments jointly created an 18-item checklist, and CLAVI Mining was the only service that cleared every item.
The company also valued customer success support that could assist the field whenever issues arose, especially because chemical plants operate 24 hours a day.
Results after implementation
EFFECT 1
[SDS/manual search time] Average 10 minutes → average 15 seconds (97% reduction). Actual measurements also improved during emergency response simulation training, with evaluators commenting that the team’s behavior had clearly changed.
EFFECT 2
[New employee safety training period] Reduced from 6 months to 3 months. The habit of asking AI on the spot whenever questions arose helped new employees become productive earlier. The burden on OJT trainers also dropped significantly, allowing them to regain time for improvement activities.
EFFECT 3
[Inquiries from overseas sites] Inquiries to the headquarters Safety and Environment Department decreased from 60 per month to 12 per month, an 80% reduction. Instant responses in local languages were a major factor.
EFFECT 4
[Near-miss incidents] In departments where CLAVI became established, the number of near-miss reports decreased, while the quality of reports, especially the precision of cause descriptions, improved and contributed to better improvement activities. The company achieved a state described as: near-miss reports decreased, but the safety culture became stronger.
Comment from the Head of Safety and Environment
The biggest concern with using AI in chemical manufacturing is that a false answer could cause an accident. CLAVI provides source evidence and transparent logs, which convinced our management team. Going forward, we want to apply it to SDS exchanges with suppliers as well.
Insights from this case and future outlook
Company D’s case shows that the higher the risk to human life in an industry, the more hallucination prevention becomes a business continuity requirement. In chemical operations, incorrect safety information can directly lead to a serious accident, so AI adoption requires explicit assurance that the technology will not provide false guidance. Meeting this requirement with general-purpose AI is practically difficult, making industry-specific services the realistic choice.
Company D started using CLAVI Mining in manufacturing sites, laboratories, and the Safety and Environment Department. It is now expanding usage to purchasing, logistics, and technical sales departments. In particular, the company is broadening application to external-facing operations such as AI analysis of SDS/MSDS documents received from suppliers and faster responses to customer technical inquiries.
In the medium to long term, the company aims to continuously incorporate changes in overseas chemical regulations such as REACH and TSCA, and to build a system that monitors regulatory compliance across the entire supply chain with AI. This aligns with the strategic theme of major chemical manufacturers: turning regulatory compliance from a cost into a competitive advantage.