Skip to content
Case Study · 01

AI-Powered Transformation of Freight Pricing Operations

How Tenfold deployed AI agents at Unotrans to compress hours of manual pricing work into minutes — and trained their team to build their own.

Unotrans

Industry

Freight Forwarding & Logistics

Headquarters

Tegucigalpa, Honduras

Engagement

Feb 2026 — Present

Service

AI Enablement Program

The Challenge

A five-person team drowning in 1,000+ pricing emails per week

1

Tariff Rate Management

Pricing officers manually collected freight rate quotes from 6–7 international cargo agents — received via email in varying formats — and entered them into Excel spreadsheets 2–3 times per week.

2

Case-by-Case Quotations

Ad-hoc quotation requests required emailing multiple agents, waiting for responses, manually analyzing each quote for cost, transit time, and free days, then recording the analysis by hand.

3

Scale of the Problem

With each officer processing 200–300 emails per week, the department handled over 1,000 pricing communications weekly — all manually. Rate validity windows of 1–2 weeks created constant pressure to refresh data.

Our Approach

The Cascade Method in action

A crawl-walk-run approach where each AI agent deployed feeds the next, creating compounding automation across the organization.

1

Crawl

Diagnose

Kickoff session bringing together stakeholders from every department. A dedicated discovery session with the pricing department followed — officers shared their screens and walked us through every step of their tariff management and quotation processes. The Asia/China tariff update cycle was selected as the first target.

2

Walk

Deploy

Within days, the team built a working proof of concept using Claude Desktop with custom skills. The first AI agent automated email extraction, data interpretation from varying agent formats, branded spreadsheet generation, and human validation — all demonstrated live within 48 hours of discovery.

3

Walk

Enable

Rather than delivering a finished solution, we conducted weekly workshops where Unotrans staff learned to create, modify, and deploy their own AI skills — with ongoing support via a shared Slack channel.

4

Run

Scale

With pricing automation established, the engagement expanded to additional departments. Coordination independently identified document reconciliation as a use case. Administration surfaced needs around invoice management and payment pattern analysis. Each new department is a new agent in the cascade.

The Solution

What the AI agent does

01

Email Extraction

Claude connects to Gmail and searches for quotation emails from the past 7 days using pricing-specific keywords — EXW, FOB, port names, carrier names, USD/EUR.

02

Data Interpretation

Claude reads each email body, interpreting structured tables, free-text responses, and agent-specific abbreviations to extract ports, carriers, rates, validity periods, and free days.

03

Spreadsheet Generation

Claude opens the Unotrans-branded Excel template, populates it with extracted data, removes empty rows, and saves a clean file.

04

Human Validation

The system pauses for the pricing officer to review and validate the output before any data is finalized, ensuring accuracy while the AI learns from corrections.

Results & Impact

Measurable impact from week one

MetricBeforeAfter
Tariff update time30–45 minUnder 5 min
Quote analysis per agent~3 min (manual)Seconds (automated)
Emails processed200–300/week/officerAI-filtered & pre-processed
Client-ready outputManual formattingAuto-generated branded Excel
Quotation responseHours to compileMinutes with AI

Non-technical pricing officers creating and modifying AI skills independently

Multiple departments proactively identifying automation opportunities without external prompting

Coordination department independently began testing Claude for document reconciliation

AI-first culture developing across the organization with collaborative learning

Started with 1 Claude Pro license — grew to a Team Plan with multi-department coverage

Planned expansion to Salesforce integration for end-to-end pricing automation

The pricing department is enthusiastic. We can see how AI can significantly reduce quotation turnaround time and cost analysis effort. The fact that the approach trains us to solve problems ourselves — rather than creating a dependency — is exactly what we needed.

— Allan Cano, Pricing Officer & Project Champion, Unotrans

Solution Architecture

Built on proven technology

AI Platform

Anthropic Claude (Team Plan)

Core AI engine for skill execution, data extraction, and document generation

Email Integration

Gmail Connector

Reads agent quotation emails, filters by keywords and date range

Data Processing

Claude Skills (Custom)

Extracts structured pricing data from unstructured email content

Output Generation

Python + openpyxl

Creates branded Excel spreadsheets with Unotrans formatting

Collaboration

Claude Projects

Shared workspace with skills, templates, and instructions

Future Integration

Salesforce

Automatic loading of validated tariff data into CRM

Timeline

From kickoff to production in weeks

Feb 9, 2026

Program announcement to Unotrans team; Slack channel created

Feb 20

Kickoff meeting — team introductions, methodology overview, first use case identified

Feb 24

Deep-dive discovery with pricing department — full process mapping

Feb 26

First demo — Claude extracting email data and generating branded Excel in real time

Feb 27

Claude 101 training — skills, projects, connectors

Mar 4

Hands-on workshop — pricing team creates skills with live guidance

Mar 6

Expanded weekly engagement; coordination department scheduled for onboarding

Mar 13

IT collaboration — Claude Code installed; Google connector integration in progress

Why It Worked

The Tenfold difference

  • Problem-first methodologyWe started with pain points, not tools.
  • Crawl-walk-run deploymentSingle-license pilot before scaling to multiple departments.
  • Enablement over dependencyEvery workshop made the team more self-sufficient.
  • Rapid proof of valueWorking demo delivered within 48 hours of discovery.
  • Embedded collaborationShared Slack channel and weekly cadence kept momentum high.

Ready to see what AI can do for your operations?

A 20-minute briefing. We’ll map your highest-impact process and show you exactly how an AI agent would handle it.