2026Agentic operations for ship chandlers
Chandler OS
An agentic layer for marine supply: quotation, commercial and operational workflows automated end to end, built on a prospecting database of Fujairah chandlers with verification and outreach-priority scoring.
Status · Built at CatalystIQ for the marine supply sector, including an enriched Fujairah chandler prospecting database.
The problem
Ship chandling runs on fast, repetitive commercial work: an RFQ arrives, items get priced from several suppliers, a quote goes back, and delivery is coordinated against a vessel that will not wait. Most of that happens in inboxes and spreadsheets, so margin leaks and quotes go out late.
The approach
- 01Build the commercial graph first: companies, vessels, contacts, verification status and outreach priority.
- 02Parse incoming RFQs into structured line items instead of reading them by hand.
- 03Automate quotation and the operational follow-through around it.
- 04Prioritise prospecting by fit and confidence rather than by list order.
What I did
- Built and enriched the Fujairah chandler prospecting database — companies, websites, contacts, verification and confidence levels, outreach-prioritisation data.
- Designed the agentic workflows over quotation and operations.
- Implemented the document and RFQ parsing layer.
Trade-offs I would defend
Structure before automation
The database and entity model came before any agent. Automating on top of unresolved, unverified records just produces confident mistakes faster.
Stack & evidence
- Python
- LLM agents
- Postgres
- Automation
- Next.js
- Role
- Founder / engineer
- Domain
- Marine supply, Fujairah
Next case study
AI QS & Planning Workbench
Construction intelligence