Industry

Digital, software and AI for logistics and freight

One of the fastest-moving sectors for AI adoption, and one where most of the value is still locked in email attachments.

Discuss your situation All services

Logistics is among the sectors making the largest year-on-year leaps in AI adoption, and the reason is unglamorous: the industry runs on documents. Booking confirmations, packing lists, customs declarations, proof of delivery — most of it arriving as PDFs and scans that somebody re-keys.

That makes it unusually well suited to the narrow class of problem machine learning is genuinely good at, and unusually badly served by generic software. A forwarder does not need a better CRM as much as it needs the CRM, the TMS, the carrier and the customs system to stop being four separate places the same shipment is described.

Why this sector is moving now

Margins in freight forwarding are thin enough that per-shipment administrative cost is a real determinant of profitability, not an overhead. A process that takes eight minutes of a coordinator’s time per booking is a cost that scales linearly with growth, which means volume erodes margin unless the process changes.

Customer expectation has shifted decisively toward visibility. Shippers who can see a parcel in real time expect the same of a container, and a forwarder who answers status questions by phone is spending coordinator time on something a system should do while competitors do not.

Regulatory reporting has tightened in parallel — customs digitalisation, goods movement declarations in several EU markets, and sanctions screening obligations that now reach well beyond the largest operators. Each of those is a system requirement rather than an administrative one.

The pressures behind it

Per-shipment admin cost: Thin margins mean coordinator minutes per booking are a profitability lever rather than an overhead.
Visibility expectations: Shippers expect container status to behave like parcel tracking, and answering by phone consumes the wrong people.
Document volume: Bookings, packing lists, declarations and proof of delivery arriving as PDFs that somebody re-keys.
Customs digitalisation: National platforms and goods movement reporting that a manual process satisfies and an automated one must integrate with.
Carrier rate volatility: Rates that move fast enough that a quote built on last month’s sheet is a margin risk.
Coordinator retention: Experienced operators are hard to replace, and the ones you have are spending their day on data entry.

Where the work usually starts

The systems and channels a logistics operators typically needs joined up. Most engagements begin at one spoke and widen only if it earns it.

Usually document extraction, because it is where the volume is and the return is calculable before anything is built. Pulling structured data out of booking confirmations and packing lists removes the highest-frequency clerical task in the building and produces a baseline anyone can measure.

After that, the sequence is normally carrier integration for status, then a customer-facing visibility portal. Building the portal first surfaces status data that has not yet been made reliable, which damages exactly the trust it was meant to build.

Marketing and brand for logistics operators

Brand Strategy & Development: Forwarders almost all describe themselves as reliable, global and customer-focused, which means none of them are differentiated. Positioning work here usually lands on a specific trade lane, commodity type or service level where the operator is genuinely better, and narrowing to it is uncomfortable and effective.
Brand Management: Consistency matters less on brochures than on the documents customers actually see: quotes, booking confirmations, arrival notices. Those are generated by systems rather than designers, which makes brand work here largely a template and integration exercise.
Social Media Strategy: LinkedIn, and a genuine argument for abandoning everything else. What performs is trade-lane commentary and capacity insight from someone who actually knows the market, which is a scarce and defensible content position in a sector full of stock photographs of containers.
Social Media Management: The constraint is subject matter access rather than production. A weekly conversation with an operations manager produces more usable content than a month of agency ideation, and it is the cadence that keeps a logistics account from becoming generic.
Content Creation & Creative Production: Photography of your own operation — yard, warehouse, handling — outperforms stock imagery so consistently that it is close to a rule. Every forwarder website uses the same container port photograph, and the ones that do not are immediately more credible.
Digital Marketing: Buying cycles are relationship-heavy and tender-driven, with long dormant periods that are not lost deals. Attribution has to account for a shipper who researched for a year and then moved when their incumbent failed once, which last-click reporting will attribute to whatever they clicked that week.
Paid Advertising: Search intent in freight is specific and low volume — a lane, a commodity, a service type — and broad terms are dominated by directories and marketplaces you cannot outbid. Exact-match on lanes and specialist handling is where a forwarder can compete.
Search Engine Optimisation: Lane and service pages are the asset: a page per origin-destination pair or per commodity type, with real transit times and requirements rather than filler. This is one of the few sectors where genuinely useful operational detail is also the best-ranking content.
Email, SMS & WhatsApp Marketing: The transactional layer matters more than campaigns. Arrival notices, delay notifications and document requests sent automatically remove inbound status calls, and in many markets — particularly the Gulf and southern Europe — WhatsApp is where shippers actually want them.
Lead Generation & Prospecting: Importer and exporter data is genuinely available from trade and customs sources in several markets, which makes targeted prospecting possible in a way it is not in most sectors. The constraint is lawful use of that data rather than obtaining it, and that has to be established first.

IT, software and AI for logistics operators

Website Design & Development: A forwarder site has two jobs: convince a shipper you handle their lane, and let them ask for a rate without a phone call. Most do neither, offering a company history and a contact form. A quote request that captures lane, commodity and volume is worth more than everything else on the site.
CRM & Sales Systems: Pipeline stages have to reflect tender cycles with long dormant periods, or the forecast will systematically overstate. Logging the incumbent and the contract renewal date matters more here than in most sectors, because that date is when the opportunity actually exists.
Business Process Automation: The core of most logistics engagements. Booking confirmation parsing, document generation, carrier status polling, invoice reconciliation — high-frequency, rule-based work with an easily measured baseline in coordinator minutes per shipment.
AI Automation Systems: Document extraction is the clearest fit in any sector we work in: unstructured input, bounded consequence, a human already checking. Confidence thresholds route uncertain documents to a person rather than guessing, which is what makes it deployable rather than a demonstration.
AI Knowledge Bases & RAG: Tariff classifications, dangerous goods rules, customer-specific handling instructions and years of correspondence. A grounded assistant with citations over that corpus answers the questions that currently route to whichever coordinator has been there longest.
AI Voice & Customer Communication: Shipment status calls are the textbook bounded use case — high volume, one question, an answer a system already holds. Anything touching customs classification or liability should escalate immediately, because a confident wrong answer there has commercial consequences.
Custom Software & Platforms: Justified where the operating model genuinely differs from what a TMS assumes — unusual consolidation, project cargo, a niche compliance regime. Not justified as a TMS replacement, and we will make that argument before quoting anything.
Data Engineering & BI: Margin per shipment by lane, customer and carrier is the report most forwarders cannot produce and most need. It requires reconciling operational and financial data that currently live apart, which is the work rather than the dashboard.
Cloud, DevOps & Infrastructure: Status integrations are customer-visible, which changes the reliability requirement. Monitoring, alerting and a tested rollback matter more here than in a sector where a failure is discovered internally.
Systems Integration: TMS to CRM, carrier APIs to status, customs platform to documents, everything to accounting. Idempotency is not optional: a duplicated booking or a double-filed declaration is a commercial and regulatory problem rather than an untidy record.
Digital Transformation Consulting: The audit usually finds coordinator time concentrated in three or four document tasks nobody had quantified. Ranking those by minutes per month, rather than by how irritating they are, changes the order of the roadmap in almost every engagement.
Maintenance & Ongoing Support: Carrier APIs change without much notice and customs platforms change on legislative timetables. An integration nobody monitors will fail quietly and be discovered by a customer, which is why a response window matters more here than a lower monthly fee.

What is specific to this sector

Customs digitalisation is uneven across the markets a forwarder serves, and that unevenness is the design constraint. Romania requires goods movement declarations through e-Transport, Italy and Hungary run their own reporting regimes, and an integration built for one does not transfer to another.

Sanctions and denied-party screening obligations now reach operators well below the size that used to attract them, and screening at the point of booking rather than at dispatch is both cheaper and safer. Building that check into the booking flow is straightforward; retro-fitting it after an incident is not.

Where customs or trade data is used for prospecting, the lawful basis needs establishing before the pipeline is built. It is genuinely available in several jurisdictions and its permitted uses vary, which is a question for a local adviser rather than an assumption.

Not legal or regulatory advice. Sector rules described here are scoping context, current to our latest review. Confirm what applies to your business with a qualified adviser.

Questions

Do we need to replace our TMS?

Very rarely, and we would argue against it. Most logistics engagements are about connecting the TMS to carriers, customs, customers and accounting. Replacing a system that currently moves freight is disruptive in a way that is hard to justify against integration.

Is document extraction actually reliable?

On structured commercial documents, yes, with confidence thresholds routing uncertain cases to a person. We measure it against a real set of your own documents before deployment and report the accuracy figure, because a number quoted before that would be invented.

Where does the return actually come from?

Coordinator minutes per shipment, almost always. It is measurable, it scales with volume, and it is the cost that quietly caps how much you can grow without hiring proportionally.

Our customers ask for visibility — what does that take?

Reliable carrier status first, then a portal. Building the portal before the status data is trustworthy publishes your own data quality problem to customers, which is worse than the phone calls it was meant to replace.

Can you help with customs system integration?

Yes, and it varies significantly by country. Which national platform applies, and what it requires, is one of the first things we establish in scoping rather than during build.

How do we compete with the digital forwarders?

Rarely on price or platform. Usually on lanes, commodities and handling where scale is a disadvantage, plus removing the administrative friction that makes a traditional forwarder feel slower than it actually is.

What does it cost?

Quoted per phase after a discovery call. Document automation in particular is one of the few areas where the return can be modelled honestly before you commit, and we would rather show that arithmetic than assert it.