Declaration volumes are up 7.9% and the number of declarant businesses is down 13.7%. The paperwork grew; the people did not. We build AI into the admin that surrounds your freight - orders, declarations, invoicing, reporting - so your team spends its week on exceptions instead of re-keying. Measured in hours back per person, per week.
Logistics contributes £175 billion in GVA and employs 8% of the UK workforce - around 7 million people making, selling and moving goods. It is also one of the most document-dense sectors in the economy. HMRC cleared 91.3 million customs declarations in 2025, 7.9% more than the year before, while the declarant population contracted from 5,650 businesses to 4,880.
That arithmetic only resolves two ways. Either the sector hires into a labour market that already has a shortage, or the repetitive parts of the paperwork stop being done by hand. 52% of logistics operators say they intend to invest in data systems for order processing and inventory. Intent is not the constraint. Adoption is.
The research shows 77% of UK businesses using AI report no revenue change from it. Not because the technology underperforms, but because nobody showed the team what it should take off their plate. Licences get bought, a few people experiment, and the rest of the operation carries on cross-checking POs against packing lists at eleven o'clock at night.
Orders arrive as PDFs, emailed spreadsheets and portal exports in a dozen formats. Claude reads them, extracts the line detail, matches it against the purchase order and the packing list, and flags only what disagrees. Your team reviews the eight exceptions instead of keying the two hundred orders.
Commodity code lookups, duty and VAT calculation and preference checks run against HMRC's published trade tariff rather than the model's memory, so the answer is sourced and auditable. The declarant still signs it off - that is the point, and it is what keeps the accountability where the law puts it.
Rate cards, surcharges, demurrage and detention reconciled against the job file, with the invoice drafted and the discrepancies listed. Customer queries answered from the actual job record instead of a hunt through Outlook.
Weekly on-time performance, exception volumes by customer, margin by lane - assembled from the systems you already run, in the format your board wants, without somebody spending Friday afternoon in Excel.
Long, repetitive RFQ and tender documents drafted from your own previous submissions, service descriptions and compliance evidence, then edited by the person who owns the relationship.
We are not a strategy firm and we are not a technology reseller. We are an AI adoption, training and governance partner, and we specialise in Claude by Anthropic - specifically Claude Cowork - rather than covering every tool on the market. Depth on one platform means your workflows are built by people who know its limits, not demonstrated by people reading the documentation the night before.
Our position is that return on AI is a behaviour change problem, not a technology problem. Most logistics firms already have AI access somewhere in the business. Very few have measurable return from it, and the reason is almost always the same: the people doing the work believe the tool is there to replace them, so they do not use it properly. We deal with that directly, in the room, before we build anything.
Because Microsoft's Copilot Cowork now runs on the same Claude models, the workflows we build with your team travel across platforms. You are not locked into a single vendor's wrapper.
An AI consultant for logistics builds AI into the admin that surrounds the freight, not the freight itself. In practice that means order paperwork, data extraction from PDFs and emails, cross-matching purchase orders against invoices and packing lists, commodity code and duty lookups against HMRC's own trade tariff, invoice raising, exception reporting and customer updates. We configure the platform, co-design each workflow with the person who currently does it by hand, and stay until it is running in the live operation. The deliverable is workflows in daily use and hours returned you can point to, not a strategy document.
Because the paperwork is growing faster than the workforce handling it. HMRC cleared 91.3 million customs declarations in 2025, up 7.9% on 2024 - an extra 6.7 million declarations. Over the same period the number of declarant businesses fell 13.7%, from 5,650 to 4,880. More volume, fewer operators. That gap is closed either by hiring into a market with a known shortage, or by taking the repetitive keying and checking off the people you already have.
No, and that fear is the single biggest reason AI stalls in logistics. The work we automate is re-keying, cross-checking and lookup - the parts nobody was hired for. Your people move to the exceptions: the mismatched consignment, the query from the customer, the classification that is genuinely ambiguous. Every workflow we build has named human approval gates, because under UK law a person remains accountable for the output regardless of which system produced it.
The public benchmark for AI users is around 2.2 hours saved per week. The organisations we work with get 6-10 hours back per person per week, depending on the role. One operations team reached 8 hours per person per week once their incoming-orders workflow was co-designed properly. The difference is not the technology - it is whether adoption was structured or left to chance.
Yes, and we assume no technical background. Our clients are finance, operations, customs and administration people in firms of 5-50 staff. The training happens on your own live paperwork, not on generic examples, so the first thing someone builds is the thing that eats their Tuesday. We specialise in Claude by Anthropic and go deep on that one platform rather than broad across many.
Yes. We are based in Scotland and work with logistics, freight forwarding, customs brokerage, warehousing and transport businesses across the UK. Delivery is in person where geography allows and online where it does not - the adoption programme is designed to run either way.
Related: if the question is less about who to hire and more about getting the team to actually use it, start with AI adoption for logistics. If your team needs hands-on capability first, see Claude training for logistics teams.