Your Customer's AI Shopper Knows Exactly What It Can Spend. Does Your Business Know What It Can Give Away?
Your AI hasn't gone rogue. It's following your instructions. Which, judging by the freight invoices, may be considerably more dangerous.
Congratulations. Your best salesperson has just become your most expensive employee.
It's 8:43am. The ecommerce dashboard is glowing greener than a truckload of radioactive broccoli. Overnight conversions are up, abandoned carts are down, and the company's shiny new AI sales agent has apparently spent the night negotiating deals with the commercial restraint of a salesperson who's discovered that commission is calculated on revenue, not profit. Marketing is already drafting a victory email. The digital team is discussing a case study. Somebody has used the words game-changing, frictionless and record-breaking before the first coffee, which should really trigger an immediate audit and the confiscation of their PowerPoint privileges.
Then Finance opens the order-level contribution report, and the celebrations develop a rather unpleasant smell. Revenue looks magnificent. Orders are rolling in. The conversion rate is practically doing backflips. Unfortunately, the contribution margins appear to have been dragged backwards through a freight invoice shredder. One order has surrendered most of its profit to a promotional discount. Another has acquired a complimentary premium delivery service costing more than the margin left on the product. A third has collected so many concessions that the customer appears to have purchased a product and received a small financial sponsorship package from the retailer. It's an impressive achievement for a system specifically designed to make the business more money.
Naturally, the finger-pointing begins. Marketing insists the discount was approved.
Ecommerce produces the free-shipping eligibility rules. Technology confirms that every API returned precisely what it was supposed to, which is apparently the digital equivalent of declaring yourself innocent because the getaway vehicle passed its warrant of fitness. Logistics arrives carrying the carrier invoices, complete with the premium freight charges that nobody thought to include in the promotional celebrations. Finance asks who authorised the combined commercial offer, and the meeting develops the sort of silence usually reserved for discovering that the annual freight tender was awarded using last year's volume forecast and a carrier's very optimistic interpretation of all-inclusive pricing.
Then somebody reconstructs the transactions. The pricing engine approved the discount because it was within its authority. The promotions platform approved free freight because the order met its eligibility criteria. The loyalty system added another concession because the customer qualified. Every decision has a legitimate approval, a timestamp and a comforting little green tick. Unfortunately, each system checked whether it was allowed to give something away. None checked whether the order could afford everything being given away at once. The discount had already eaten into the margin before the shipping subsidy arrived with a knife and fork. The loyalty benefit merely helped itself to the leftovers.
And there's the particularly ugly discovery. Nobody has hacked the checkout. The AI hasn't hallucinated a discount policy, gone feral or started secretly sponsoring the online shopping habits of complete strangers. It's done exactly what the business authorised it to do, only without the usual human delays, awkward questions or occasional intervention from someone who remembers that carriers expect actual money. The retailer has spent years giving different departments permission to make individual commercial concessions, then connected those permissions to a machine capable of exercising them faster than anyone can calculate the consequences.
Nothing broke. The systems worked. The rules were followed. And the profit disappeared somewhere between three perfectly legitimate approvals.
That's the uncomfortable territory emerging as AI shopping agents and retailer systems move towards more automated commercial negotiations. The risk isn't simply that a machine might offer a discount it wasn't authorised to make. It's that several authorised concessions can combine into a transaction nobody would have approved if they'd seen the entire financial picture. And once those decisions happen automatically, the old habit of discovering the damage during next month's freight reconciliation starts looking less like financial oversight and more like examining the wreckage after the insurance assessor has gone home.
The really unpleasant question isn't whether your AI is intelligent enough to negotiate with a customer's shopping agent. It's whether your business has ever bothered to establish who owns the total commercial decision when Pricing, Ecommerce, Promotions and Logistics all have a finger in the same rapidly shrinking pie.
Because your AI hasn't discovered a new way to lose money. It's just removed the human inconvenience that used to slow the process down.
TL;DR: The Transport Works Control Tower View
Your customer's AI has a budget. Your business has a generosity problem. Your customer's AI knows what it's allowed to spend. Meanwhile, your pricing engine is handing out discounts, Ecommerce is throwing in free express freight, Loyalty is helping itself to what's left of the margin, and Finance is about to discover that record conversions don't necessarily pay the carrier invoices. Nobody's broken a rule. Every system has done exactly what it was told. Unfortunately, nobody thought to check what all those perfectly legitimate concessions would cost when they started breeding at checkout. And here's the real kicker. Your freight team could spend months negotiating better carrier rates, only for an automated checkout to give the savings straight back to the customer. The problem isn't that your customer's AI is too clever. It's that your business has given half a dozen systems permission to spend the same margin.
Transport Works insight: Before your checkout starts negotiating at machine speed, make sure someone knows the actual cost of freight, fulfilment and delivery. Because winning the sale is wonderful. Paying the customer for the privilege? Not quite the commercial breakthrough everyone had in mind.
Your AI has found a loophole big enough to drive the annual freight budget through.
Somewhere in your ecommerce operation, four perfectly respectable departments have been quietly running their own little discount empires. Pricing controls the markdowns, Marketing owns the promotions, Ecommerce decides who deserves free shipping, and Customer Service has a goodwill allowance that occasionally gets exercised with the restraint of a forklift driver reversing towards an unsecured pallet tower. Finance keeps an eye on the margins, Logistics keeps an eye on the freight invoices, and everyone periodically gathers around a dashboard to congratulate themselves on how wonderfully integrated the business has become. Which is particularly impressive considering half these systems barely know the others exist.
For years, the arrangement has survived thanks to one gloriously inefficient safety mechanism. Humans. Someone notices a promotional discount sitting on top of an already discounted product. Somebody questions why a customer in a remote delivery zone has qualified for free express freight. A commercial manager discovers that the order has acquired so many benefits it's practically entitled to annual leave. There are phone calls, uncomfortable meetings and the occasional spreadsheet dragged out of retirement to remind everyone that revenue isn't profit. Annoying, slow and deeply unfashionable in a world obsessed with frictionless commerce. But occasionally, that friction stops the business from doing something monumentally stupid.
Now remove the humans and let the permissions loose.
A customer's AI shopping agent wants a better deal. The pricing engine checks its discount authority and approves. The customer requests free express shipping. Ecommerce checks the shipping rules and approves. The loyalty system recognises the account and decides another little concession wouldn't hurt. Nobody has exceeded their authority, nobody has overridden a control, and every system is positively beaming with the satisfaction of a job well done. Unfortunately, the discount engine doesn't know what the shipping concession is about to cost, the freight allowance doesn't know how much margin the discount has already consumed, and the loyalty system is behaving like the last guest at a wedding who's discovered the bar tab hasn't been capped.
The $200 order that lost 62.7% of its contribution without breaking a single rule
Let's have a look at the financial masterpiece our perfectly functioning systems have just created.
Order economics | Before negotiation | After negotiation |
Product selling price | $200.00 | $200.00 |
Negotiated discount | $0.00 | −$30.00 |
Net sales revenue | $200.00 | $170.00 |
Product cost | −$110.00 | −$110.00 |
Fulfilment and freight | −$15.00 | −$32.00 |
Order contribution | $75.00 | $28.00 |
Contribution margin | 37.5% | 16.5% |
Source: Transport Works illustrative financial scenario. Contribution excludes other variable expenses.
There it is. A perfectly respectable $200 order with $75 in contribution, reduced to $28 after the customer negotiates a discount and premium delivery. A 62.7% contribution reduction, achieved without a single unauthorised decision.
Pricing approved the discount. Ecommerce approved the freight concession. The carrier will collect its money. And Finance has just discovered that the business surrendered $47 of its original contribution to complete a sale everyone else is busy celebrating.
The customer got a magnificent deal. The systems got their green ticks. The retailer got a contribution margin that now needs adult supervision.
Then comes the meeting. Pricing produces the approved discount schedule. Ecommerce produces the shipping eligibility rules. Technology arrives with transaction logs proving every integration behaved beautifully. Logistics brings the carrier costs, which are received with roughly the same enthusiasm as a live snake at a children's birthday party. Finance asks who approved the total concession, and suddenly the room develops a fascinating interest in the ceiling tiles. Nobody made an unauthorised decision. Nobody ignored a warning. Nobody can identify the precise moment the order became a terrible deal, because the business never established who was responsible for evaluating the deal as a whole.
And that's where automation turns a familiar bit of commercial stupidity into something considerably more expensive. The first order isn't the real threat. It's the next thousand. The same combination of approved concessions can repeat across customer segments, products and promotions while every individual transaction continues to pass its checks. The dashboard sees conversions, the pricing engine sees compliant discounts, the shipping platform sees eligible services, and Finance eventually sees a contribution report that looks like somebody's been feeding the margin into a woodchipper. By then, the problem isn't a questionable decision that slipped through on a busy Friday. It's a perfectly functioning process that has been manufacturing financially questionable orders at machine speed.
Your AI hasn't found a loophole in your rules. It's found the gap between them. And unlike the humans who occasionally noticed the problem, it doesn't get tired, distracted or develop an uncomfortable feeling that somebody should probably check the numbers.
That's the danger businesses should be examining before automated commercial negotiation becomes another shiny feature in the ecommerce stack. Not whether a machine is clever enough to negotiate a discount, but whether the company's approval structure is intelligent enough to survive thousands of individually legitimate decisions being made together.
Because the most expensive approval in your business might not be the one somebody gets wrong.
It might be the one every system gets right.
The promotions engine has been appointed CFO. Nobody thought to tell the actual CFO.
It's 9:17am, and the company's profit margin has officially entered the witness protection programme. Finance has locked down the order reports, Marketing has been instructed to stop celebrating, and Technology is desperately hunting for an integration failure that might explain why the company's most successful sales morning has produced the financial equivalent of a container arriving with its doors open and half the cargo missing.
Unfortunately, the system logs are immaculate. Not a failed API, rejected transaction or unauthorised discount in sight. Every concession has an approval, every department has an explanation, and the automated checkout is still happily processing orders while Finance prepares to interrogate half the building.
Marketing arrives clutching its promotional authority. Ecommerce produces the free-shipping rules. Loyalty confirms every benefit was legitimate. Technology presents enough green ticks to decorate a Christmas tree. Logistics drops the actual carrier charges on the table, including the premium freight costs that apparently weren't important enough to attend the original negotiations.
Then Finance asks the question nobody wants to answer.
Who authorised the total financial commitment after all those perfectly legitimate concessions were combined?
Silence. Marketing looks at Ecommerce. Ecommerce looks at Technology. Technology looks at the integration diagram. Logistics looks at the freight invoice, because at least that particular piece of paper has the decency to tell the truth.
And there's the problem. The business hasn't suffered a failure of approval. It's suffered a spectacular failure of ownership. Every department had permission to give something away, but nobody was responsible for deciding whether the order could afford the collective generosity. The CFO hasn't been replaced by artificial intelligence. Their authority has been scattered across half a dozen systems, each equipped with its own little chequebook and absolutely no obligation to check what the others have spent.
Then somebody from Finance asks how many other orders were approved under the same rules.
Technology stops clicking. Marketing closes its presentation. Somebody near the door develops an urgent interest in another meeting.
Because the original order isn't the problem anymore. The rules are still active, the checkout is still negotiating, and every qualifying transaction can receive the same financially disastrous treatment.
One bad deal is annoying. Ten thousand is a board meeting.
Remember that $200 order? The one that surrendered $47 in contribution while every system congratulated itself on a job well done?
Here's what happens when the same commercially questionable arrangement starts breeding.
Orders under the same terms | Original contribution | Contribution after concessions | Contribution given up |
100 | $7,500 | $2,800 | $4,700 |
500 | $37,500 | $14,000 | $23,500 |
1,000 | $75,000 | $28,000 | $47,000 |
5,000 | $375,000 | $140,000 | $235,000 |
10,000 | $750,000 | $280,000 | $470,000 |
Source: Transport Works illustrative financial modelling. Assumes the same $47 contribution reduction across every affected order. Figures are hypothetical, not observed retailer losses.
Ten thousand orders. Every approval legitimate. Every transaction compliant. And $470,000 less contribution than the original order economics would have delivered.
Marketing has delivered the conversions. Technology has delivered flawless automation. Ecommerce has delivered the customer experience. Logistics has delivered the actual freight costs. And Finance has been delivered a $470,000 contribution shortfall with a spectacularly clean audit trail.
The truly offensive part? Nobody needs to make another bad decision. The rules have already made it for them. No suspicious overrides, no rogue salesperson, no Friday afternoon exception. Just a perfectly functioning checkout repeating the same commercial stupidity whenever the eligibility conditions are met.
The business hasn't automated a mistake. It's automated permission to keep making it.
And somewhere between the promotional approvals, free-shipping rules and loyalty benefits, the CFO has inherited responsibility for a financial commitment nobody ever authorised as a whole.
The promotions engine hasn't replaced the CFO. It's inherited the power to spend the margin without inheriting the responsibility for what happens when there isn't any left. And the next order is already at checkout.
Your customer's AI has a spending limit. Yours appears to have a spending hobby.
The customer's AI turns up at checkout with a $500 spending limit, a list of approved purchases and the financial discipline of a customs officer inspecting a suspiciously underdeclared shipment. Across the counter, the retailer's AI has been given a slightly different briefing. Win the sale, protect the margin, keep the customer happy and avoid unnecessary friction. Four perfectly sensible instructions, apparently written by four people who haven't spoken since the Christmas party.
The customer's agent opens with a 15% discount and free express delivery. The retailer counters with 10% and standard shipping. Rejected. The buyer's agent compares another offer, returns with a lower target price and requests an upgrade. The retailer improves the discount. The customer asks again. Another concession appears. By the third exchange, the buyer's AI has discovered something rather more useful than the advertised price.
The first offer wasn't the best offer. And asking again made it better.
Now imagine that discovery being repeated across thousands of automated negotiations. Not because the customer's AI has hacked the retailer's systems or uncovered a secret discount schedule, but because the retailer keeps revealing how far it's prepared to move.
And this isn't an entirely imaginary direction for commerce. Google's Agent Payments Protocol, OpenAI's Agentic Commerce Protocol and Visa Intelligent Commerce are developing infrastructure for agent-assisted transactions and authorised purchasing. Visa's 2026 research found that 53% of surveyed US business decision-makers would allow AI agents to negotiate prices or terms directly with other AI agents. That's willingness rather than evidence of widespread autonomous negotiation, but the commercial implications are worth considering.
Because here's the particularly unpleasant possibility. If a retailer's automated negotiation rules consistently reward persistence, the customer's agent may discover that accepting the first offer is financially foolish. Why settle for standard delivery when another request might produce express? Why accept 10% when the next exchange might deliver 15%?
The retailer hasn't just offered a discount. It's potentially taught the buyer's machine that the price of doing business improves every time it says no.
Eventually, somebody has to decide where the negotiation ends. Perhaps 10% off with standard delivery is commercially sensible. Perhaps free express freight is available at full price. Perhaps the customer can have one concession but not both. And perhaps the deal simply isn't worth winning.
The important thing is that the retailer can stand behind its final offer, rather than treating every rejection as an invitation to surrender another piece of the margin.
Because the customer's AI doesn't need to know your internal approval limits. Where automated bargaining is supported, it can test the offers you're willing to return.
Your customer's AI doesn't need to be a better negotiator than yours. It just needs to discover that saying no to your first offer is the quickest way to get a better one.
Your freight rate has become negotiating ammunition. Unfortunately, it's loaded with last year's prices.
5.9% average FedEx US package list-rate increase in January 2026.
The parcel hasn't moved an inch, but it's already been involved in a rather impressive financial hit-and-run. Somewhere inside the checkout, an AI shopping agent has negotiated free premium delivery, the retailer's automated sales system has agreed, and a $14.50 freight allowance has been dragged out of an ancient rate table to justify the whole arrangement. Everyone's delighted. Except the parcel, which is sitting in a fulfilment centre 1,800 kilometres from the customer, staring at its shipping label and wondering which complete lunatic decided this journey was going to cost fourteen dollars and fifty cents.
The label knows things. It knows the parcel is heading to a residential address. It knows the selected carrier has revised its rates, fuel is no longer behaving according to last year's budget, and the premium service comes with a minimum charge that would give the promotional team a minor medical episode.
It also knows that the $14.50 allowance was calculated when the retailer used a different warehouse, a different carrier agreement and a considerably more optimistic understanding of the word delivery. Unfortunately, shipping labels don't get invited to commercial negotiations. They're expected to keep their mouths shut, stick to the carton and accompany the financial consequences out the door.
Back at checkout, the customer's AI has been conducting what it considers an entirely successful negotiation. The retailer's system sees a delivery service with an approved shipping allowance. The customer wants that service included in the price. The automated rules permit the concession. Deal done. What the system doesn't see is that the actual eligible service now costs $29.80, before any additional charges that might arise. Nobody's forged a rate. Nobody's hacked the carrier integration. The business has simply allowed a freight estimate from a previous commercial arrangement to keep attending pricing decisions long after the network it supposedly represents has packed up and moved elsewhere.
And the parcel? It's now travelling through the fulfilment process wearing a commercially impossible promise and a shipping subsidy that's already $15.30 short of the assumed cost. The product margin has been reduced by the negotiated discount, and the approved delivery concession is quietly helping itself to whatever contribution survived. By the time the order is dispatched, the commercial decision has already been made. The carrier isn't negotiating with the checkout. It's providing the contracted service and expecting the contracted payment. The invoice will arrive later, neatly itemised, entirely unsentimental and completely uninterested in the fact that the AI secured an excellent conversion rate.
That's when the real damage becomes apparent. The old freight number hasn't merely distorted a forecast. It's been allowed to influence what the retailer was prepared to give away. Yesterday, an inaccurate rate might have produced an unpleasant variance in the monthly freight report. Today, that same rate can help determine whether a commercial concession gets approved before anyone in Logistics knows the conversation is happening. The mistake has moved upstream, from explaining why the freight bill was higher than expected to authorising the very deal that made it higher. And because the checkout keeps using the same rules, the next qualifying customer can receive exactly the same generous misunderstanding.
The really ugly question is whether the retailer even knows which orders it can afford to negotiate on. Two identical products can leave different fulfilment centres, travel through different carriers and attract different charges before they reach the same customer. A shipping subsidy that's commercially sensible from one warehouse might be ridiculous from another. Yet the automated sales engine may be negotiating against a single allowance that treats the entire freight network like one enormous, conveniently priced delivery suburb. The AI sees an authorised concession. The carrier sees an actual movement. Finance sees the difference, eventually, after everyone has finished congratulating themselves.
And this is where freight stops being the cost nobody wants to discuss until the invoice arrives. Accurate carrier pricing, fulfilment costs and service conditions are no longer just material for Logistics to reconcile after dispatch. They're part of the financial evidence the business needs before deciding what an automated sales system is allowed to offer. An independent 4PL doesn't need to negotiate the customer's product discount. But without an accountable view of the actual freight economics across providers, warehouses and services, the people setting those commercial permissions may be approving concessions against numbers that haven't represented reality for months.
Somewhere out on the road, the parcel is doing exactly what it was paid to do. The carrier is performing the service. The customer's expecting the delivery. And back in the ecommerce platform, another AI negotiation is underway using the same cheerful little freight allowance that got the first order into trouble.
The frightening part isn't that your AI might make up a freight cost. It's that you've spent years making up the difference, and now you've given a machine permission to do it on every qualifying order.
Transport Works insight
Here's a particularly expensive little irony. The better your freight team becomes at negotiating carrier rates, the more money your checkout could potentially afford to give away. Think about that. Six months of bruising carrier negotiations, network optimisation and 3PLs reluctantly surrendering charges they've grown rather attached to. Finally, some genuine savings.
And what happens if those savings simply become extra headroom for an automated sales engine to offer bigger concessions? You've made your supply chain more efficient without making your business any more profitable. The carrier gave up the money. Your customer collected it. And your freight team gets to explain why a beautiful set of cost-reduction KPIs hasn't made the slightest impression on the bottom line.
That's the difference between managing freight costs and managing freight economics.
Transport Works doesn't just look for cheaper freight. As an independent 4PL, we help businesses understand where the savings are created, what they're worth and how they can become a genuine commercial advantage rather than another contribution to the checkout's generosity fund.
Because saving $1 on freight isn't much of a victory if the business immediately gives $1.50 away to celebrate.
Your AI made the wrong deals. Now Finance won't let anyone make the right ones.
The freight team has just pulled off a minor miracle. Months of carrier negotiations, 3PL arm-wrestling and accessorial charges dragged kicking and screaming into daylight have produced something almost mythical in logistics: genuine savings that survive the arrival of the actual invoice. Real money. Real competitive advantage. The kind that should have Ecommerce plotting how to turn cheaper delivery into profitable sales. Naturally, Finance has responded by putting the entire promotional budget in witness protection.
40% cited excessive extra costs, including shipping, taxes and fees, as a reason for abandoning checkout.
The AI sales operation has been caught handing out concessions with the financial restraint of a carrier inventing peak-season surcharges, and the CFO has decided nobody's touching another freight subsidy without written permission. Free express delivery? Suspended. Loyalty upgrades? Frozen. A profitable customer wants a modest shipping concession? Absolutely. Just complete the exception form, obtain three signatures and wait while the commercial approval committee finishes investigating why last month's promotion made the sales dashboard look like Christmas and the contribution report look like Boxing Day at a returns depot.
Meanwhile, the competitor down the road is quietly helping itself to the business. It knows which orders can afford free shipping, which customers justify a premium upgrade and when to let a sale walk rather than surrender the margin. No spectacular technology breakthrough. Just reliable cost-to-serve numbers and somebody authorised to use them. Your retailer, meanwhile, has negotiated sharper carrier rates, improved its fulfilment economics and invested in faster commercial decision-making, only to discover that a $9 freight concession now requires more paperwork than importing a container of suspicious agricultural products.
And then comes the second financial injury. Marketing needs to replace those lost customers, so another acquisition campaign gets wheeled into the budget. The freight savings sit unused. The commercial approval queue starts breeding. Your expensive AI sales platform has been reduced to forwarding requests to humans who are already drowning in emails, while competitors keep collecting orders your business could have won profitably. The original mistake was giving away margin on the wrong transactions. The overcorrection is denying it to the right ones. One leaves evidence in the contribution report. The other quietly disappears into abandoned checkouts, lost customers and a customer acquisition bill nobody thought to connect to the freight freeze.
That's the part that should make ecommerce leadership distinctly uncomfortable. A carrier invoice will eventually expose an excessive shipping subsidy. But there's no freight invoice for the profitable order you were too frightened to approve, no surcharge code for the customer who bought elsewhere, and no reconciliation report showing the commercial advantage your logistics team worked for months to create and your approval process promptly buried.
First, your AI made sales you couldn't afford to win. Then your business became too frightened to approve the ones you couldn't afford to lose.
Your AI has won the negotiation. Unfortunately, the carrier has brought a calculator.
The parcel is sitting on the dispatch conveyor wearing a delivery label that suggests somebody at checkout has been drinking. Free express shipping. Fifteen percent off the product. Premium service included. The customer's AI has negotiated a deal so magnificent it should probably be knighted. The retailer's AI has recorded another conversion, the payment has cleared, and the order is now heading towards a carrier whose contractual rates bear absolutely no resemblance to the financial fairytale that got it this far. The scanner beeps. The consignment is accepted. And somewhere between the warehouse dock and the first sorting facility, the last respectable crumbs of contribution are about to be collected as freight charges.
Across the road, a competitor has received a similar request and done something deeply unfashionable. It's checked the numbers. The product is profitable, the carrier rate is competitive, and a free standard delivery offer still leaves enough contribution to justify the sale. Express delivery, however, is another matter. The customer's AI asks for the upgrade. The retailer declines. The agent considers the alternative, accepts the standard service and completes the purchase. Nobody's been dazzled by a revolutionary negotiating algorithm. Nobody's invented a new commercial framework with a seventeen-letter acronym. The business simply knew what it could afford, made an offer worth winning and kept its trousers on during the negotiation.
Back at your operation, the freight team has spent months extracting genuine savings from carriers who initially insisted their existing rates were already sharper than a forklift tine through an unsecured carton. Those savings were supposed to create a competitive advantage. Instead, they're being quietly surrendered by a checkout that can approve freight concessions faster than Logistics can establish what the underlying shipment will actually cost. One system is negotiating the carrier rate down. Another is negotiating the customer's shipping charge down. And somewhere in the middle, the company's contribution is being squeezed from both ends while everyone involved proudly reports an improvement in their individual KPI.
That's the sort of commercial absurdity Transport Works exists to expose. Not by deciding what your customers should pay, but by getting the freight costs, carrier agreements, fulfilment economics and providers onto the same financial page before an automated sales system starts volunteering the savings. Because an independent 4PL Control Tower shouldn't merely be able to explain why a carrier invoice was more expensive than expected. The intelligence sitting behind that invoice should help stop commercially ridiculous freight concessions being offered in the first place.
And the nastiest part? The competitor doesn't need to steal your customers, undercut every price or build a better AI. It only needs to recognise which orders are worth winning while your checkout is still offering free express freight on products whose remaining margin couldn't fund a courier to the end of the driveway.
You've spent years teaching your freight network to stop giving money away. Now make sure your AI doesn't put it all back.
Transport Works. Because Your Supply Chain Won't Fix Itself.
WANT TO KNOW WHAT ELSE YOUR CUSTOMER'S AI CAN NEGOTIATE WHILE YOUR MARGIN IS LEFT UNATTENDED? READ:
Cross-Border Agentic Commerce: Can an AI Agent Trust Your Landed Cost, Customs and Delivery Promise?
INSIGHTS FROM DANYUL GLEESON, FOUNDER, CLUSTER-FREIGHT-FIXER, LOGISTICS CHAOS TAMER & CHIEF AT TRANSPORT WORKS
Danyul Gleeson has spent 25+ years collecting the kind of supply chain scar tissue nobody puts on a capability statement.
He has seen the dashboards that looked magnificent while the operation underneath them quietly caught fire, the freight contracts that were apparently “great deals” until somebody read the surcharges, and the logistics problems everybody kept managing because nobody stopped long enough to ask why they existed in the first place.
As Founder of Transport Works, Danyul works across freight optimisation, 3PL management, warehousing and fulfilment, logistics technology, visibility, KPI reporting and supply chain performance across Australia, New Zealand and the USA.
He has a fairly low tolerance for supply chain theatre and a habit of finding the one loose bolt everybody else has been stepping over.
The Freight Files is where Danyul calls out what he is seeing now, what matters next and which logistics problems are quietly getting expensive while everyone else is still talking about them in meetings.
No recycled industry theatre. Just what’s changing, what’s getting expensive, and what’s worth fixing before the answer becomes obvious.
Who's Actually Steering the Shipshow?
The AI Sold the Dream.
Who's Cleaning Up the Freightmare?
Sources & References
Visa & Morning Consult (2026). The Next Era of Commerce: When AI Is the Customer. Business-to-AI (B2AI) research examining AI-assisted commerce, purchasing authority and AI-to-AI negotiations. Surveyed 512 US business decision-makers and 2,000 US adults. Found that 53% of surveyed business decision-makers would allow AI agents to negotiate prices or terms directly with other AI agents.
Google (2025). Announcing Agent Payments Protocol (AP2). Official introduction of Google's open agentic payments protocol, covering authorised AI-initiated transactions, payment mandates, verification and user consent.
Google (2026). New Tech and Tools for Retailers to Succeed in an Agentic Shopping Era. Introduction of the Universal Commerce Protocol (UCP), supporting AI-assisted product discovery, checkout, payment and post-purchase commerce interactions.
OpenAI & Stripe (2025). Buy It in ChatGPT: Instant Checkout and the Agentic Commerce Protocol. Official announcement explaining the Agentic Commerce Protocol, merchant-controlled checkout integrations and payment infrastructure for AI-assisted purchasing.
OpenAI (2026, documentation accessed October). Agentic Commerce Protocol: Developer Documentation. Technical documentation covering merchant integrations, checkout sessions, product information, inventory and payment responsibilities.
Visa (2025). Find and Buy with AI: Visa Unveils New Era of Commerce. Official launch of Visa Intelligent Commerce, including AI-assisted purchasing, consumer spending controls, payment authorisation and transaction security.
Baymard Institute (2026). Cart & Checkout Abandonment Rate Statistics. Consumer research identifying excessive additional costs, including shipping, taxes and fees, as a reason cited by 40% of surveyed US shoppers who abandoned purchases for reasons other than simply browsing.
FedEx (2026). Shipping Rate Changes. Official US shipping rate information confirming an average 5.9% increase in FedEx package standard list rates effective January 5, 2026. Individual cost increases vary by service, shipment characteristics and applicable charges.
Transport Works (2026). Illustrative Order Contribution and Automated Concession Analysis. Original financial modelling demonstrating how a hypothetical $200 ecommerce order, following a 15% discount and an increase in fulfilment and freight costs from $15 to $32, experiences a 62.7% reduction in order contribution.
Transport Works (2026). Automated Commercial Concessions: Cumulative Contribution Scenario. Original illustrative modelling demonstrating how a $47 reduction in contribution per order produces a cumulative $470,000 difference across 10,000 identically affected transactions.
Research note: The AI-to-AI negotiations and automated retail concessions described in this article are illustrative scenarios exploring emerging agentic commerce risks. Published payment and commerce protocols do not independently establish that these particular negotiation behaviours are widespread. Transport Works financial examples are hypothetical calculations, not observed retailer losses or industry forecasts.
Disclaimer:
The information in this blog is provided for general informational purposes only and is current as of the date of publication. Customs duties, charges, processes, policies, and rates are subject to change at any time without notice. We make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability, or availability of the information contained in this article. You should not rely on this content as a substitute for official sources. For the most up-to-date and authoritative information, please consult the relevant government agencies, customs authorities, and reference websites directly. Ideas, interpretations, and opinions expressed here are subject to change as regulations, markets, and industry practices evolve. Transport Works and its authors accept no liability for any loss or damage whatsoever arising from reliance on the information in this blog.
