
Logistics

Seven Billion's dispatch optimization solver turned container placement from a reactive scramble into a predictable margin driver.

Michael Stangel
Founder & CEO
Lummid Containers

About
A wholesale container supplier unified scattered listings into an AI-searchable marketplace.
Industry
Logistics
Founded
2000
Business Context
Business context and operating model
One of North America's largest wholesale container networks The client, a North American container supplier, has supplied new, used, and specialty shipping containers, including reefers, flat racks, open-tops, and military-specific equipment, for more than two decades, operating out of over 350 container depots and terminals across all 50 states. Purchasing had always run through a sales rep: a customer would call or email, a rep would check availability across a patchwork of depot spreadsheets, PDFs, and inboxes, and a quote would come back hours or days later.
Problem Definition & Framework
Current State
Failure Modes
Decision Criteria
KPI Baseline
Problem definition and framework
An inventory that existed, but wasn't searchable The equipment customers wanted was almost always in the network somewhere — the problem was that no single system knew where. Depot inventory lived in independent spreadsheets, condition reports arrived as scanned PDFs, pricing changed by hand in email threads, and container specs were scattered across hundreds of documents in inconsistent formats with no shared schema. A customer looking for "a 40ft high cube, wind-and-watertight, near the Gulf Coast, under a specific budget" had no way to ask that question directly. Instead, every search became a manual relay through a sales rep who had to piece the answer together from memory and a handful of spreadsheets, which meant slow answers, inconsistent quotes, and lost sales to customers who simply gave up and went elsewhere.
Decision framed: How can buyers search available containers by need, condition, location, price, and specification when source data is scattered across depots, PDFs, spreadsheets, and emails?
Framework used: Unstructured catalog extraction and natural-language inventory retrieval
Baseline mapped: current process, failure modes, handoffs, data owners, and decision latency.
Success criteria tied to business KPIs, adoption behavior, and operating cadence changes.
How it works
Before / After

Architecture

Process Flow

Data Pipeline

Results

Delivery Timeline

Engagement Method & Process Diagram
01 Discovery
02 Data Audit
03 Baseline
04 Prototype
05 Pilot
06 Production
07 KPI Monitoring
Engagement method and process diagram
Process diagram: Depot spreadsheets + PDFs + emails + inspection notes -> extraction and deduplication -> structured container catalog -> semantic search index
Engagement method: Source mapping; schema design for container specs; extraction pipeline; search relevance testing; marketplace workflow integration.
Delivery pattern: discovery, data audit, baseline measurement, prototype, pilot workflow, production rollout, and KPI monitoring.
A natural-language marketplace built on top of a unified, always-current catalog We built an extraction pipeline that continuously ingests listings, specs, condition reports, and pricing from hundreds of unstructured sources — depot spreadsheets, scanned PDFs, dealer emails, and inspection notes — and normalises them into a single structured, de-duplicated catalog that stays in near real-time sync with what is actually sitting at each depot. On top of that catalog sits a natural-language search experience: customers log in, describe what they need in plain English — container type, condition, location, price range — and get matched results ranked by fit and proximity.
Architecture, Results & Measurement
Solution architecture, results, and measurement
Measurement loop: compare weekly operating behavior against baseline, surface exceptions, and review model/workflow drift with business owners.
KEEP READING
Explore more studies on AI, analytics, and enterprise intelligence.










