A freight forwarder with a simple promise, anything from anywhere to anywhere, was generating around 300 leads a month and answering only as many as six people could by hand. A third were never touched. FlowX.AI built an agentic lead-management application: two AI stacks that capture, deduplicate, classify and score every lead before it reaches a human, an automatic WhatsApp first touch within minutes, a quote calculator that generates the offer PDF, and a dashboard that makes follow-up discipline visible for the first time.
Projected results, validated pre-production
What if the third of your pipeline nobody ever calls turned into revenue, without spending more on ads?
Demand was never the problem. Lead ads generated around 300 leads a month, and a six-person team, four in sales and two running everything else, answered as many as six people could. All client communication ran through a single business WhatsApp account, phone numbers copy-pasted by hand. Each agent kept profitability in a private Excel with formulas per transport type, and the final quote reached the client as a screenshot of a spreadsheet. Behind the hundred untouched leads sat roughly 96 quotes a month that went cold after a single touch, and eighteen hours per agent per week of copy-paste, WhatsApp and Excel administration.
Leads arrived faster than a six-person team could work them by hand. Three things defined the starting state.
In freight forwarding, conversion is slow: the client only thinking about a first import signs six months later, but only if someone stays in touch. New leads arrived, older interested ones were forgotten, and nothing made forgetting impossible.
The team raised the ad budget to grow volume. Cost per lead tripled with no additional leads. The diagnosis was unglamorous: the problem was not generation, it was that a third of everything generated was never touched by anyone.
Who did their follow-ups today? How many quotes went out this week? A six-person team, half working from home, no dashboard, sales tasks in a shared Excel with no reminders. One belief, that some leads got an automatic message, turned out in discovery to be a myth: the platform notified an operator, who replied by hand.
The team no longer copies phone numbers or reconstructs quotes from screenshots. Each pressure now has a concrete answer, built to work with the channels and spreadsheets that already existed.
A third of monthly leads never contacted, and around 96 quotes a month going cold after a single touch.
Automatic capture of every lead into a controlled workflow per contact, an automatic WhatsApp first touch within minutes of submission, and a follow-up agent that fires a predefined message after a set interval. Follow-up statuses cannot be saved without a due date.
All client communication through one personal business WhatsApp, phone numbers copy-pasted by hand.
The WhatsApp Cloud API on the existing business account, with an approved message template. FlowX configured the ad-platform business and developer side end to end.
Profitability formulas living in agents’ private spreadsheets; quotes sent as screenshots.
A quote calculator inside the application with dynamic fields per transport type (ocean LCL, rail, air express, air cargo) that generates a branded offer PDF, with formulas ingested from the agents’ own Excels and verified figure by figure against them.
Two agents unknowingly calling the same client, with no visibility into response time or follow-up discipline.
Assignment by click, the first agent to claim a lead owns it, plus a management dashboard with daily cards, a funnel by status, per-agent performance and a live priority queue flagging anything uncontacted for 12 hours or overdue by three.
Personal data (names, phone numbers) flowing into an AI-scored pipeline.
Personal data tokenized before it reaches the model, and scripted deduplication on every batch pulled from the ad platform.
Success was defined before the build, in numbers the client can hold over the system, on three commitments.
An intake stack governs what reaches a salesperson; a specialised stack turns raw data into a prioritized, quoted, followed-up action. No task reaches an agent without passing through both.
Ingest every lead
The Lead Ingestion Agent captures each lead from the ad platform and starts a controlled workflow per contact, so nothing arrives without a record.
Read the intent
The Channel Intent Classifier separates a new lead from a follow-up and from an existing client re-inquiring, so the wrong task is never created.
Check the pipeline
The Duplicate Contact Verifier checks against the live pipeline before any new entry, so two agents never call the same client.
Score & place
The Lead Qualification Agent scores each lead and places a prioritized task on the dashboard. The guarantee the stack exists to make: no lead lost, duplicated or misrouted.
Rank by client value
The Priority Scoring Calculator ranks urgency on weights agreed live in discovery (existing client, volume, weight), surfaced as one to three stars. The business logic belongs to the client, not the model: the recurring customer outranks the one-off.
Quote and send
The Quote & Offer Tracker builds the quote in the calculator, attaches the offer PDF, sends it over WhatsApp, logs it, and starts the follow-up timer.
Close the loop
The Automated Quote Follow-Up Agent fires after a set interval, closing the loop on the roughly 96 quotes a month that used to die in memory and goodwill.
Brief before the call
The Lead Summary Extractor writes a concise brief before the first call, so no agent dials blind.
A lead never disappears now. The ones we lost, we have to admit it, new ones came in and the interested ones in the back we forgot along the way. What I will finally see is who is not doing their follow-ups, and that is the most important thing.
Built kick-off to UAT with real data in six weeks, around three weeks of active implementation, scoped to work with what existed rather than waiting for anyone to change behavior.
Contract & kick-off
Contract signed; verticals, team and timeline agreed.
Discovery, on-site
On-site discovery and design review. The end-to-end journey and statuses agreed, the WhatsApp automation myth dismantled, and a scope expansion added the quote calculator.
Build
Flow, statuses, AI prioritization, dashboard and quote calculator built. The ad-platform integration went live during build, with real leads flowing in and notifications suspended.
UAT & pilot go-live
UAT on real data, with the ocean LCL formula validated live against the client’s own spreadsheet. Pilot go-live, into a six-month pilot with weekly feedback and monthly business reviews.
The application was scoped to work with what existed: the WhatsApp business account already in use, the agents’ own pricing spreadsheets, and a six-person team’s workflow. The ad-platform integration ran live during the build, with real leads flowing in and notifications suspended.
One distinction was made honestly in discovery: the application working perfectly and the volume arriving are two different things. Functionality and marketing performance are measured separately, on weekly sales feedback and monthly business reviews for the duration of the pilot.
Three lenses on the same result: the coverage and speed it captures, the discipline it makes visible, and the foundation it lays for the rest of the roadmap.
Two-thirds of leads used to be contacted; the target is all of them, the single figure that carries the business case on its own.
Around 96 quotes a month that used to go cold after a single touch now sit on a timer that fires on its own.
Profitability that lived in personal spreadsheets and client history that lived in memory become a shared, queryable base.
On the surface this is a lead-management project: a light CRM with WhatsApp automation for a six-person team. Underneath, it is the beginning of a data infrastructure for a company that had none. Profitability per shipment lived in personal spreadsheets, invisible to management; client history lived in people’s memories. From go-live, every quote, margin and interaction becomes structured data.
That is what makes the rest of the roadmap real rather than aspirational: a quoting agent that pulls rates from carrier APIs, an invoice-reconciliation agent that matches the four to six supplier invoices per shipment against the client invoice and flags unprofitable jobs before they close, and a customs-clearance agent that pre-validates documents before they reach the broker.
At depth, the story is about turning response time into a competitive weapon. In freight forwarding the client asking for a quote asks three companies at once, and the winner is usually not the cheapest, it is the first to answer competently. The system compresses every link in that chain: a message within minutes instead of a call the next day, prioritization that pushes the valuable client to the top, a quote and a PDF generated inside the application instead of a spreadsheet and a screenshot, and follow-up with a due date instead of memory. Discipline stops being optional, because it becomes visible.
We are the enterprise orchestration platform that enables regulated institutions to deploy deterministic, zero-hallucination agentic workflows on top of legacy infrastructure. Logistics and freight-forwarding operators run critical customer and operations journeys on FlowX.AI, including:
Bring a value stream and we will show you a path to production: every lead captured, a first touch in minutes, quotes generated in the app, and follow-up you can finally see.