A regional agricultural cooperative in the U.S. Midwest came to Armakuni for aI-powered chemical safety monitoring and conversational chatbot for chemical mixing operations.
The challenge
Armakuni engaged A regional agricultural cooperative in the U.S. Midwest to prove that AI could make chemical mixing operations safer, using AWS funding to cover the full cost of the proof of concept. The POC established the technology and built the foundation the production monitoring system will run on. A separate engagement in March and April 2025 covers the cooperative's security posture, with external penetration testing and an internal vulnerability assessment.
Scale: Agricultural cooperative managing 142 chemical Safety Data Sheets across field and mixing operations.
What we built
Armakuni delivered aI-powered chemical safety monitoring and conversational chatbot for chemical mixing operations. The engagement ran as fixed-scope POC: Proactive Monitoring and Chatbot Interface (January 21, 2025). Separate fixed-scope SOW: External Penetration Testing and Internal Vulnerability Assessment (January 21, 2025). AWS funded the work at $10,000.
The outcomes
5 measurable outcomes shipped across the engagement. The ones that moved the business the most:
A $10,000 engagement cost the cooperative $0, fully covered by AWS funding. The POC was structured to qualify for AWS investment, so the cooperative validated the technology and built the production foundation without spending anything to do it.
142 Safety Data Sheets moved from manual files to a database the system queries automatically. Every piece of chemical safety information is now instantly accessible to the monitoring tool, the chatbot, and any future capability, available in milliseconds instead of a worker stopping to search.
All three AI capabilities work on real data from the cooperative's own operations. Computer vision reads the actual mixing program screen, the chatbot answers from the verified SDS database, and the knowledge base is built from the cooperative's own 142 sheets, none of it synthetic.
The infrastructure built for the POC is what the production system builds on. There is no throwaway prototype to replace, so reaching full production means adding capabilities to the running, secured, two-data-center system rather than starting over.
A defined production roadmap sits on architecture that is already validated. Real-time desktop monitoring, barcode scanner integration, and acreage-based quantity calculations are all scoped onto the architecture the POC proved, with no need to go back to basics.
Built on AWS
The production environment runs on 11 first-party AWS services. Delivered under our AWS AI Services competencies.
What's next
Production build: real-time desktop monitoring on Windows machines, barcode scanner integration for automated chemical identification at the point of mixing, acreage-based quantity calculations, and full integration with the existing mixing software.