Skip to main content
Email Intelligence / Quote AutomationLive within the ORCA group

AI Email Intelligence and Quote Automation for a National Freight Company

A national freight and logistics provider with over 100 staff was drowning in email. Thousands of messages daily across sales, operations, and customer service — with critical quote requests buried in the noise. We built a two-phase AI system: intelligent email management first, then a RAG-powered quote engine that turns historical pricing data into instant freight quotes.

100+

Staff supported

70%

Faster quote turnaround

1000s

Emails processed daily

2-phase

Deployment approach

The Problem

Freight logistics runs on email. Rate requests, booking confirmations, pickup schedules, delivery updates, claims, compliance documents — all flowing through shared inboxes that multiple staff members monitor simultaneously. At this company's scale, the email volume created three distinct problems.

Problem 1: Email triage consumed hours every day. The sales team spent the first 60–90 minutes of each day sorting through overnight emails, trying to identify which messages needed immediate action (a rate request from a key account) versus which could wait (a routine delivery confirmation). With multiple staff watching the same inboxes, work was frequently duplicated — two people responding to the same enquiry, or worse, nobody responding because each assumed someone else had handled it.

Problem 2: Quote requests were the bottleneck. When a customer or prospect requested a freight quote, the process required a sales rep to manually search through historical quotes, find comparable shipments (similar weight, route, cargo type), adjust pricing for current fuel surcharges and market conditions, compose the quote, get internal approval for anything above a threshold, and send it back. A single quote could take 30–60 minutes. During busy periods, quote turnaround stretched to 24–48 hours — and in freight, slow quotes lose deals.

Problem 3: Institutional knowledge walked out the door.The most experienced sales reps could quote from memory because they'd seen thousands of similar shipments. But when someone left or was on leave, that knowledge went with them. Junior staff had to start from scratch on every quote, often calling colleagues for guidance or making conservative (high) estimates that cost the company competitive bids.

Phase 1: Email Intelligence Layer

The first deployment focused on taming the email chaos. We built an AI layer that sits between the company's Microsoft 365 environment and their staff, intelligently routing and pre-processing every inbound message.

Intelligent Classification

Every inbound email is classified by intent: rate request, booking confirmation, delivery update, claim, compliance document, general enquiry, or internal communication. The AI reads the full email body (not just subject lines) and understands context — a message saying “can you do Sydney to Melbourne, 2 pallets, next Tuesday?” is correctly tagged as a rate request even without the word “quote” appearing anywhere.

Smart Routing & De-duplication

Classified emails are routed to the appropriate team member based on account ownership, expertise, and current workload. If a message comes from a key account, it goes to their dedicated rep. If that rep is unavailable, it routes to the next qualified person. The system also detects when the same request arrives through multiple channels (forwarded emails, CC chains) and consolidates them — eliminating the duplicate-response problem entirely.

Draft Assist

For routine responses (delivery ETAs, booking confirmations, document requests), the system generates draft replies using the company's tone, terminology, and policies. Staff review and send with a single click. For complex messages, the system provides a suggested response structure and pulls relevant context — previous correspondence with that customer, recent shipment history, any open issues — so the staff member starts with full context rather than searching for it.

Phase 2: RAG-Powered Quote Engine

The second phase tackled the quoting bottleneck with a retrieval-augmented generation (RAG) system that gives every sales rep access to the company's full quoting history — turning years of accumulated pricing data into an instant competitive advantage.

How the RAG Quote System Works

The system ingests the company's historical quote database — thousands of past quotes with route, cargo type, weight, pricing, margins, and outcome (won/lost). This data is indexed using vector embeddings that capture the semantic similarity between shipments, not just exact matches.

When a new quote request arrives, the AI extracts the key parameters (origin, destination, cargo type, weight, dimensions, urgency) and searches the historical database for the most similar past quotes. It doesn't just find exact route matches — it understands that a Sydney-to-Brisbane pharmaceutical shipment at 500kg is comparable to a Sydney-to-Gold-Coast medical equipment shipment at 450kg.

The system then generates a draft quote with recommended pricing, showing the comparable historical quotes that informed the recommendation, the current fuel surcharge adjustment, and a confidence indicator. The sales rep sees exactly why the system recommends a particular price — not a black box, but a transparent recommendation backed by data.

Win/Loss Learning

When quotes are won or lost, that outcome feeds back into the system. Over time, the pricing recommendations become more competitive — the AI learns which price points win deals in specific corridors and adjusts recommendations accordingly. This is the institutional knowledge preservation that the company was missing. When a senior rep retires, their decades of pricing intuition remain embedded in the system.

Results

Quote turnaround reduced by 70% — from an average of 45 minutes per quote to under 15 minutes. Complex multi-leg shipments that previously took hours now take 20–30 minutes.

Email triage time eliminated — the 60–90 minutes staff spent each morning sorting emails is now handled automatically. Staff start their day with a prioritised, pre-classified inbox.

Zero duplicate responses — the de-duplication system ensures exactly one person handles each request, eliminating the confusion and professional embarrassment of multiple responses to the same customer.

Junior staff quote like veterans — the RAG system gives every sales rep access to the company's full pricing history. New hires produce competitive, well-informed quotes from day one instead of after months of accumulated experience.

Quote win rate improved — pricing recommendations informed by historical win/loss data produce more competitive quotes without sacrificing margin. The system identifies the pricing sweet spot for each corridor and cargo type.

Technology Stack

N8N (workflow orchestration)
Claude AI (classification & drafting)
Microsoft Graph API (email)
Supabase + pgvector (RAG store)
Vector embeddings (semantic search)
Custom quote database

Why Logistics Is Ripe for AI

Freight logistics is one of the most email-intensive industries in existence. Every shipment generates a chain of communications — booking, confirmation, pickup notification, in-transit updates, delivery confirmation, proof of delivery, invoicing. Multiply that by hundreds of shipments per day across a national network, and you get an operation that is simultaneously critical and overwhelmingly manual.

Most logistics companies have already digitised their operations with TMS (transport management systems), but the connective tissue between systems — the emails, the quote calculations, the customer communications — remains stubbornly manual. AI fills exactly this gap: not replacing the TMS, but handling the coordination layer that sits on top of it.

The RAG approach is particularly powerful in logistics because pricing is contextual, not formulaic. A route that costs X per kilogram for general freight might cost 2X for hazardous materials and 0.7X for a high-volume regular customer. Human reps carry this context intuitively. A RAG system makes it explicit and accessible to everyone.

Key Lessons

Deploy in phases — prove value before expanding scope. Phase 1 (email intelligence) delivered immediate, visible ROI that built organisational trust. When Phase 2 (RAG quoting) was introduced, staff were already comfortable with the AI handling part of their workflow. Trying to deploy both simultaneously would have created change management resistance.

Transparency in AI recommendations is non-negotiable for sales teams.Sales reps won't trust a price that comes from a black box. The quote engine shows exactly which historical quotes informed each recommendation and why. This transparency converted sceptics into advocates — reps started actively contributing feedback to improve the system.

Historical data is an undervalued asset. This company had years of quoting data sitting in spreadsheets and email archives. By structuring and indexing it, we turned a passive record into an active competitive advantage. Most companies have similar untapped data — the challenge is extraction and structuring, not the AI itself.

Frequently Asked Questions

How can AI help a freight or logistics company?

+
Three high-ROI areas: email intelligence (classifying, routing, and drafting replies to the hundreds of emails per day that logistics operations receive), quote automation (generating freight quotes from historical pricing using retrieval-augmented generation), and operational monitoring (flagging at-risk shipments, tracking ETAs, proactively notifying customers of delays). Logistics is one of the highest-email-volume industries, which is exactly what modern LLMs are best at.

What is a RAG quote engine and why is it useful for logistics?

+
RAG (Retrieval-Augmented Generation) is a technique where an AI retrieves relevant historical data — in this case, past quotes for similar shipments — before generating a new quote. For logistics, this means the AI can produce a draft quote in seconds using the same pricing logic your team has applied over years of operations, rather than making up a price from scratch. Your estimators review and adjust the draft rather than building each quote from zero. Typical time savings: 70%.

How accurate is AI email classification for freight and transport enquiries?

+
For logistics-specific email classification (new quote, booking change, POD request, invoice query, general enquiry), accuracy typically lands between 94% and 98% once the system is tuned with 200–500 labelled examples from your own inbox. Routing errors are cheap to recover from because misrouted emails still end up visible — just in the wrong queue. The system also learns from every manual re-routing.

Do I need to change our existing TMS or freight software?

+
No. AI email intelligence and quote automation sit alongside your existing transport management system — they read, write, and trigger actions via API or email. We have integrated with CargoWise, MyToll, GoodShip, and custom in-house TMS platforms. Your operators keep working in the tools they know while the AI handles the upstream email and quoting workload.

What does an AI system for logistics cost in Australia?

+
A phased rollout typically runs $20,000–$45,000 AUD for the build and $300–$800 AUD per month for infrastructure. Phase 1 (email intelligence + draft assist) delivers meaningful ROI within 30 days. Phase 2 (RAG quote engine) needs 6–12 weeks of historical quote data to reach best-in-class accuracy but usually pays back within the first quarter.

Running a logistics operation?

If your team is spending hours on email triage and manual quoting, we can build the same system for your operation. Book a call and we'll walk through the ROI.

Book a Free Call →