Will AI agents place wholesale orders? Scenarios for B2B fashion buyers
Forecasts say AI agents will intermediate large volumes of B2B buying. For fashion wholesale, where orders are seasonal and relationship-led, the realistic path is narrower and slower.
KEY TAKEAWAYS Summary by the editors
- As reported by Digital Commerce 360, Gartner predicts AI agents will intermediate more than 15 trillion US dollars of B2B spending by 2028, on the condition that they can rely on verifiable data feeds and standardised trust frameworks.
- The Gartner figures cover B2B purchasing in general and do not break out fashion wholesale, where buying is seasonal, assortment-led and relationship-driven.
- The most plausible near-term agent tasks in fashion wholesale are replenishment of core lines, availability checks and order status queries, not first-time collection selection.
- Wholesale suppliers need machine-readable catalogues, price lists, minimums and stock data before an agent can transact reliably.
- Verification of agent identity is a prerequisite for B2B ordering, and signed-request methods such as Web Bot Auth already exist for that purpose.
Will AI agents really place wholesale orders for fashion buyers?
Some will, but in a narrow set of situations and later than forecasts for B2B in general suggest. Digital Commerce 360 reported Gartner's prediction that AI agents will intermediate more than 15 trillion US dollars of B2B spending by 2028, and that agents could handle a large majority of B2B purchases within three years, in both cases conditional on verifiable data feeds, standardised trust frameworks and automated exchanges. These are forecasts about B2B as a whole, mostly in sectors with standardised goods.
Fashion wholesale differs in ways that matter. Buyers choose from a seasonal collection they often need to see and touch, orders are placed months before delivery, assortments are built around store clientele and prices and minimums are negotiated within relationships. An agent has less to automate here than in the purchase of office supplies or components.
Which wholesale tasks are most suited to agents?
The tasks that look automatable are the repetitive, rules-based ones. These include re-ordering core styles that sell steadily, checking availability and delivery dates, tracking order status and flagging price or stock changes. The tasks that depend on judgement, such as selecting a new season's range or deciding how deep to buy a risky fashion item, are the least suited.
| Task | Suitability | Why | Data required |
|---|---|---|---|
| Re-order of core and continuity lines | Higher | Rules-based, repeat demand | Live stock, price lists, order history |
| Availability and delivery date checks | Higher | Factual look-up | Real-time inventory and delivery windows |
| Order status and invoice queries | Higher | Structured information | Order and shipment events |
| Price and minimum comparison across suppliers | Medium | Terms differ by account | Account-specific price rules |
| Seasonal collection selection | Lower | Taste, brand fit and appointments | Imagery, fit and merchandising context |
| Negotiating terms and delivery changes | Lower | Relationship and exceptions | Contract data and approval rules |
What would suppliers need before an agent can order?
An agent can only act on information it can read and trust. For a supplier that means catalogue data with consistent identifiers, size and colour breakdowns, prices by customer group, order minimums and multiples, and stock and delivery information that is current. Gartner's own condition, as reported, is that agents rely on verifiable data feeds, which implies that stale or inconsistent data is a barrier, not an inconvenience.
Many wholesale suppliers hold this information in separate systems that do not agree: an ERP for stock, a product information tool for descriptions and a spreadsheet for price lists. Agents would expose those gaps quickly, because they will not guess when a field is missing.
Assortment context is a further gap. Wholesale buyers decide using imagery, fit information, delivery windows, price positioning and the brand's story, not only attribute fields. An agent that compares two suppliers' jackets using a few fields will miss what a buyer would see in a showroom or an appointment. Brands should expect agents to be used for narrowing and checking, and should keep investing in the human and visual parts of selling that agents cannot replace.
How would a supplier know the buyer's agent is legitimate?
Identity is the first gate. In consumer payments, card networks and infrastructure providers have introduced methods to verify agents. Digital Applied reports that Visa's Trusted Agent Protocol, announced in October 2025 with Cloudflare, uses HTTP message signatures so merchants can verify that an agent is genuine. Cloudflare documents Web Bot Auth, in which a bot signs requests with a key whose public half is published on its own domain.
A B2B supplier would need an equivalent for buyer agents, tied to a customer account, an approved credit limit and a named person who authorised the agent. Without that chain, a supplier cannot tell an authorised procurement agent from a script.
What scenarios are plausible for the next few years?
- Assistive. Buyers use AI tools to compare line sheets, summarise terms and prepare orders, but a person submits every order. This is the most likely path in fashion wholesale.
- Delegated replenishment. A buyer authorises an agent to re-order named core lines within set limits from approved suppliers, with exceptions sent to a human.
- Supplier-side agents. Brands deploy assistants that answer buyer questions on availability and terms, while orders still go through the existing portal.
- Open agent buying. Agents from retailers or marketplaces search across brands and place orders with minimal supervision. This needs identity, liability and data standards that did not exist as settled practice in 2026.
These are editorial scenarios, not predictions with probabilities. Their order reflects how much trust infrastructure each requires.
Who is liable if an agent orders the wrong goods?
Worldpay reports that in consumer card payments no liability shift exists for agentic transactions and that disputes short of fraud are still being negotiated. B2B payment terms differ, with invoicing and credit common in wholesale, but the same question applies: who bears the cost of an order the buyer says they did not intend. Suppliers should expect to write the answer into trading terms, including which orders an agent may place, order value caps and how cancellations are handled.
What should a wholesale brand do now?
- Make catalogue, price and stock data consistent and accessible through an API or structured export.
- Document order rules: minimums, multiples, delivery windows and customer-specific terms.
- Decide which tasks, such as re-order and status checks, you would let a verified agent do.
- Write agent-related clauses for trading terms with legal advice.
- Ask key accounts how they use AI tools today, so planning rests on observed behaviour, not on forecasts.
Sales teams should be part of the planning. Reps know which accounts re-order predictably, which negotiate every season and which already use digital tools heavily. That knowledge identifies where an assisted or delegated re-order flow could save time without damaging the relationship, and where it would be unwelcome. Starting with two or three willing accounts and measuring order accuracy and time saved is more informative than building for a market that does not yet exist.
Frequently asked questions
Will AI agents replace wholesale buyers in fashion?
There is no evidence of that. Forecasts such as Gartner's concern B2B purchasing broadly, and fashion buying depends on taste, seasonality and relationships. Agents are more likely to take over repetitive tasks such as re-orders and status checks.
What does Gartner predict about AI agents in B2B buying?
As reported by Digital Commerce 360, Gartner predicts agents will intermediate more than 15 trillion US dollars of B2B spending by 2028, conditional on verifiable data feeds and standardised trust frameworks.
What do wholesale suppliers need to prepare for agent orders?
Consistent product identifiers, size and colour breakdowns, customer-specific prices, minimums, live stock and delivery data, plus a way to verify the agent and the person who authorised it.
How can a supplier verify an AI agent?
Signed-request methods exist. Cloudflare documents Web Bot Auth, where bots sign HTTP requests with a key published on their own domain, and Visa announced a Trusted Agent Protocol built on similar signatures for consumer payments.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
