Skip to content
← Back to Our Work

ChemMasters

Order and email automation with a closed learning loop for a chemicals distributor

ChemMasters architecture schematic

One example subsystem from this build. In most engagements we deliver the entire stack top to bottom — application code, Docker, infrastructure, deployment, and scaling — all in-house. Where a client already has platform teams in place, we work alongside them.

Project Summary

Production AI platform for a chemicals distributor — two integrated halves. (1) Email automation: a Celery Beat dispatcher polls inbound mail every 60s, acquires PG advisory locks per email for crash-safe mutual exclusion, hands off to a Unified Email Agent (Claude) that classifies, extracts, and routes; a companion Email Response Agent drafts replies. A truth-set eval harness keyed by Sage OR# regression-tests every Claude prompt change against curated real-world cases before promotion. (2) Data platform: unifies legacy Sage product/customer data with semantic search (Vertex AI Search), AI trip-report generation, and a document pipeline that extracts SDS/TDS PDFs via GCP Document AI. MCP server and REST APIs expose the unified data to AI agents and a Slack/React surface.

Our role: Architecture & Engineering

What We Did

  • Customer emails classified, extracted and routed into the legacy Sage ERP, with a person approving each order before it posts
  • Every human correction captured as a Knowledge Bank lesson and a regression truth case
  • Playbooks promoted only on measured telemetry
  • Nightly accuracy audits against real order history
  • Legacy Sage product and customer data unified behind semantic search and an MCP server
  • SDS and TDS document pipeline built on GCP Document AI

Want this in your own workflow?

Tell us which process you'd start with. We map it, install the loop, and prove it before it widens.

Plan Your Rollout