AI Sales Agent for B2B & Wholesale

The buyer reality

Generic ecommerce chatbots weren’t built for this.

Most storefront chat assumes a consumer with a cart: one unit, list price, ship tomorrow. A wholesale buyer arrives with a quantity, a spec sheet and a deadline — and a search box gives them a product grid, while a generic chatbot gives them a paragraph. Both lose the enquiry your sales team never saw.

Availability

“Is 500 units available, and when?”

The first question in almost every bulk enquiry, and the one a product page answers worst. Stock is shown for one unit; lead time for 500 lives in someone’s inbox.

MOQ & tiers

“What’s the MOQ, and the price at 5,000?”

Minimums and tier breaks are the real price list, and buyers bounce when they can’t find them. Negotiated account pricing makes it worse — the honest answer is often “we quote.”

Specs & tolerance

“Does it meet this exact spec?”

In technical wholesale, a small tolerance difference doesn’t downgrade the product — it rejects it. A near-miss answer here costs a return, a claim, or the relationship.

Logistics

“What does it cost landed, palletised, at my port?”

Export buyers think in freight, incoterms and container fill — questions no product page answers and no consumer chatbot understands.

What the agent does

Answer what the catalog can. Capture the rest as an RFQ.

HOLLOU is catalog-grounded: it answers only from your own site and catalog data — specs, stock, MOQs, published tier pricing, shipping policy. When the answer isn’t in the catalog, it is instructed to say so instead of guessing, and with your lead form on the buyer can leave the enquiry for your team.

Grounded answers

Spec, compatibility, substitution and availability questions answered from your catalog, with the values taken from retrieved records rather than the model’s general knowledge.

RFQ capture

Anything quote-shaped — volume pricing, lead times, freight — becomes a captured lead — the buyer’s contact details and the question they were on — ready for your sales desk.

Recommendation, in spec

When the exact item is out of stock or discontinued, it proposes alternates that meet the same spec from your own catalog — and is instructed to say so plainly when none do.

Checkout when it fits

For ready-to-ship SKUs it still adds to cart and hands to your checkout. For everything else, the RFQ is the conversion.

A wholesale buyer, worked through

A modeled example — not customer results

“Can you do 500 of the 40mm stainless flange couplings, and how fast?”

The agent checks the catalog: 320 in stock now, MOQ 100, and the listing carries a stated production lead time for larger runs. It answers with those numbers, and offers to raise an RFQ for a confirmed date on the full 500 — captured as a lead with the buyer’s contact details attached.

“What’s the unit price at 1,000 versus 5,000?”

If tier pricing is published in the catalog, it quotes the tiers as listed. Here it isn’t — so the agent says pricing at that volume is quoted per order, and turns the question into a quote request with both quantities on it, rather than inventing a price.

“I need the gasket in EPDM, rated to 120°C, tolerance ±0.05mm. Which SKUs qualify?”

It filters on the specification tables and returns only exact matches — two SKUs. A third comes close on temperature but misses the tolerance, and the agent is instructed to say so rather than sell around it.

“Can you ship a pallet to Hamburg, and what would freight run?”

Shipping policy in the catalog confirms the destination is served. Freight cost isn’t in the catalog, so the agent doesn’t estimate one — it captures the RFQ as a lead — contact details and the question as the buyer put it — for your team to price.

How it fits

One line on your site. Your process stays yours.

The agent embeds with a single script tag on any site — a Shopify or WooCommerce storefront, a custom B2B portal, a marketplace-style listing site, or an export seller’s catalog pages. It works alongside your existing store and quote workflow rather than replacing them: checkout stays on your platform, and RFQs land with your sales team. It’s the same agent we run for other high-consideration catalogs, tuned for quote-driven buying — and plans follow catalog size and volume, not seats.

Questions

What B2B merchants ask us.

Most of our orders end in a quote, not a cart. Does this still fit?

That’s the case it’s built for. The agent treats the RFQ as the conversion, not a fallback: it answers what the catalog can, then captures the rest as a lead with the buyer’s contact details and the question they were on. Cart and checkout are there for the SKUs that are ready to ship.

Where do captured RFQs go?

Into your merchant console as leads — the buyer’s contact details and the conversation context — where your sales team picks them up. On Pro and above, transcripts show personal details masked by default. Nothing is quoted or committed on your behalf; the agent gathers, your team closes.

Our real prices are negotiated per account. What does the agent tell buyers?

Only what your catalog publishes. If tier pricing is listed, it quotes the tiers; if pricing is quote-only, it says exactly that and captures the request. It is instructed not to invent a number, and it holds no account-specific or negotiated pricing at all — it knows only what your catalog publishes — so there is no private rate for it to leak.

Does it work for marketplace-style and export sellers?

Yes. It embeds on any site with one script tag, so marketplace-style catalogs and export sellers’ own sites both work. For export, the buyer questions that matter — MOQs, production lead times, ports and incoterms — are answered where the catalog covers them and captured as RFQs where it doesn’t.

What does a pilot actually involve?

A 30–60 day pilot on the pages you choose, with success criteria agreed in writing before launch and no long-term contract. If it doesn’t show what we both hoped, it ends, and removing the agent is one line in your template.

Next step

See it answer questions from your own catalog.

Send us your store URL. We’ll index a slice of your catalog and show you the agent answering real questions about your actual products — before you commit to anything.

No catalog credentials needed for the first preview

Apply as an early merchant Try the interactive demo