Not a chat widget. Four products on one install — a buyer agent that answers from your live catalog, a support concierge, analytics on every conversation, and a merchant intelligence layer that tells you which products your catalog can’t sell and what to fix. All live today.
Official Alibaba.com channel partner · backed by startup & cloud programs from
In the live product, every value in the answer — price, contents, stock — is retrieved from the merchant’s catalog at the moment of reply; nothing is written by the model. This preview replays those answers. Watch a recording of the real product →
For cloudy water in an 18ft round pool, two options under $100 — both in stock.
I’d start with the Green Algae Package — green algae is the most common cause of cloudiness, and it leaves budget room.
Values in annotation ink are retrieved, not generated.
Generic assistants improvise. HOLLOU reads your live catalog and shows its work — the price it quoted, the stock it checked, the product it matched. When your catalog can’t answer, it says so instead of guessing.
Most assistants end at the shopper. Ours is a loop — what it learns selling to your customers comes back to you as work you can act on.
Full-screen guided discovery grounded in your live catalog. Product cards, comparison, budget filtering, add to cart and checkout hand-off — and it says so when it doesn’t know.
For your shoppersOrder status, returns and post-purchase questions, answered from your own policies, with escalation to a human when it should be a person.
For your inboxEvery session: the funnel from engaged to cart to checkout, revenue attribution, and full transcripts with personal details masked by default.
For your decisionsThe questions your catalog couldn’t answer, which product each was about, a listing-quality score with the specific fix, and whether your coverage is improving.
For your roadmap— and one thing none of them do alone
Every other assistant’s failures are invisible — the shopper leaves and nobody learns anything. Ours become your to-do list. A bolt-on widget can’t do that, because it never knew your catalog well enough to grade it.
Guided discovery needs room for images, specs, comparison and price. That’s the whole argument for taking the screen — not a preference, a constraint.
“Something easy to set up, under $3,000…”
HOLLOU is catalog-agnostic — the qualifier isn’t the category, it’s the behavior. It earns its place wherever the basket is large, the specs matter and the shopper has real questions, and any deep catalog where the wrong choice means a return. It is not built for impulse buys.
Where it’s earning its place today —
Sizing against the yard, install effort, pump & filter compatibility, delivery access.
Fit & installation →Room fit, material trade-offs, what goes with what, the return risk of getting it wrong.
Dimensions & material →Spec matching, compatible parts and consumables, the accessories a first-timer forgets.
Compatibility →Deep catalogs where the part number matters and the wrong choice costs a return.
Part & spec matching →— and any deep catalog like
Selling something research-heavy that isn’t here? Tell us about your catalog →
Most companies at our stage show a roadmap and let you assume it’s all shipped. Here is the honest split — what you get now, and the two things we’re still building.
| Capability | Status |
|---|---|
| Buyer Agent Full-screen guided discovery grounded in your catalog, with add-to-cart and checkout handoff. | Live |
| Merchant analytics Funnel, transcripts with PII masked by default, and detection of questions your catalog couldn’t answer. | Live |
| Support Concierge Order status, returns and post-purchase questions, with escalation to a human agent. | Live |
| Merchant Intelligence Listing-quality scoring and catalog gap detection, drawn from the questions shoppers actually asked. | Live |
| Voice commerce Phone and SIP-ready product discovery for merchants who still sell over the phone. | Target Q4 2026 |
| AI visibility (AEO/GEO) Tracking and improving how your products surface inside ChatGPT, Gemini and Perplexity. | Target Q4 2026 |
If a row shows a target quarter, it isn’t in your contract and we won’t demo it as though it were. We’d rather lose a deal on scope than win one on a promise we haven’t built.
We’re pre-launch and we’re not going to pretend otherwise. What we offer early merchants is the thing you can’t buy later: direct access to the people building it, and real influence over what gets built next.
We configure it against your catalog ourselves. You don’t get handed a doc.
On a controlled slice of your traffic, not your whole storefront.
We write down what “working” means before launch, so neither side moves the goalposts.
If the pilot doesn’t show what we hoped, it ends. One line removes the agent.
Usage-based — typically $30 per 1,000 engaged shoppers — at preferential rates that hold after the pilot.
The in-development list gets reordered by what early merchants actually need.
Send us your store URL. We’ll index a slice of your catalog and show the agent answering real questions about your actual products — before you commit to anything. No catalog credentials needed for the first preview.