Hiulix
Talk to the agent
Pre-launch Free 60-day pilots — five to eight stores, run by the founder himself

The most intelligent first reply
an appliance seller can put online.

Hiulix answers the questions that decide the sale — will it fit under the cabinet, which filter belongs to this model, does it ship before Friday — from your catalog and nothing else. Then it tells you what the conversation produced. At 9pm. On a Sunday.

Talk to the agent See pricingFrom $129/mo

This is the real product, running on our own site. Bring your own store and argue with it.

app.hiulix.com/inbox
Visitor Live Intent: comparing
Hi — I’m an AI assistant, not a person. What are you shopping for?
An air fryer for a family of four. Our counter is pretty shallow though.
Then depth matters more than capacity here. How deep is the space, and does it need to sit under a wall cabinet?
About 13 inches deep, and yes — under the cabinet.
Introducing Hiulix

Most website chat was built to deflect support tickets. Appliance sellers point it at sales anyway, so it answers three FAQs, asks for an email, and calls that a lead. Meanwhile the shopper deciding between a 5-quart and an 8-quart has a real question and nobody awake to answer it — and the cheaper model next to yours on a marketplace does not need explaining to win.

Hiulix is built for one job and one trade: the conversation that decides whether a shopper buys. It answers from your catalog and nothing else, checks the thing actually fits before the order rather than after it, suggests the accessory that belongs to that model, and turns the exchange into a line you can act on. When it reaches the edge of what your catalog says, it stops and gets you — which is the part most of this category still gets wrong.

An empty desk by a window in a quiet office, laptop open

Most appliance returns were never faulty. They are expectation gaps — it did not fit, or it was not the model that takes that filter. Both are questions somebody asked, or would have asked, before buying.

0shopper questions left until morning
<5sto a first, useful answer
2models recommended instead of four hundred
An AI sales assistant for appliance sellers

Anything your best salesperson does in the first ten minutes, Hiulix does at 9pm.

Not a general e-commerce chat tool you configure into usefulness. Appliance data has a shape — capacity, wattage, voltage, dimensions, accessories, compatible model numbers, warranty terms — and the product ships knowing it, so you load a catalog and get a working assistant instead of a configuration project.

“Will the 6-quart fit under a cabinet?” Comparing
“Which filter fits model CM-220?” Compatibility
“Can I get it before Friday?” Ready to buy

It reads intent, not keywords.

Every message is classified — browsing, comparing, ready to buy, asking a price, asking wholesale terms, needing help, reporting a problem — and the conversation takes the path that matches.

Three colleagues working together at a table with laptops

It checks the thing fits before the order.

Height under the cabinet, which filter belongs to which model, whether that accessory actually attaches. A shopper who is told no before paying does not become a return, and a shopper who is told yes buys with more confidence.

Cart recovered · $219 auto
Accessory offered · basket liner auto
Owner notified · wholesale auto

It finishes the job, then gets out of the way.

Nudges the hesitating cart, offers the accessory that fits, flags the dealer enquiry, logs the whole exchange. You open a daily summary of what shoppers asked, which products drew the most questions, and which questions it could not answer.

The market, honestly

A large, mature market that is slowly moving online.

Estimates of the small-appliance market disagree by tens of billions of dollars, because each firm draws the category differently — kitchen only, or vacuums and personal care too; retail value, or manufacturer revenue. Here are five 2026 estimates side by side, rather than the single flattering one.

PM Persistence Market Research
—
SM Statista Market Insights
—
DM Deep Market Insights
—
MI Mordor Intelligence
—
VI Vyansa Intelligence
—

Base years are 2025 or 2026 and forecast horizons run 2030–2034, and the top two estimates are global while the bottom three are U.S. only — the bars compare magnitude, not identical periods or identical scopes. A sixth estimate, Research and Markets at $111.7B for 2026, is named here rather than charted because it publishes no growth rate. The U.S. figures cluster between about $10B and $15B with growth in the low-to-mid single digits. This is not a boom, and the case for Hiulix does not need one. It rests on the shift of those sales online and toward small sellers: one 2026 tracker puts about 40% of U.S. small-appliance purchases online, and industry research counts roughly 8,900 U.S. businesses selling small electrical appliances online, up about 9% in a year. Vendors in this category publish striking conversion figures — one reports 12.3% of shoppers who chat go on to buy against 3.1% of those who do not. Those are self-selected comparisons published by the companies selling the tools, so they are not repeated here as a promise. Tora Labs' own planning assumption is an 8% relative lift, and the first job of the pilot is to find out the real number.

Knowledge layer

Answers that only say what your catalog says.

The assistant answers from sources you load and can see — your product data, your manuals and spec sheets, your warranty, returns and shipping policies. Scoped to your store and never mixed with another seller's.

  • It does not guess a specification. If a wattage or a dimension is not in what you gave it, it says so and offers a handoff. A made-up spec on your site becomes a return.
  • No safety or regulatory claim it was not given. It will not say a product is UL listed, or free of anything, unless you supplied that as data.
  • Every source is switchable. Turn a page off and the agent stops quoting it on the next message.
  • One store, one catalog. Nothing you upload is shared across sellers, and you can export or delete it.
  • Model-independent by design. The orchestration layer sits between your business and whichever model runs underneath, so the provider can change without the product changing.
The sales console

Every conversation becomes a line you can act on.

The same question means different things at different moments. Each message is sorted into one of seven paths and run through a pipeline from conversation to lead to quote to order. It sounds like this: “Which one suits a small kitchen?” → comparing

Browsinganswer
Comparingrecommend
Ready to buycapture
Reporting a problemescalate

Seven paths, one conversation.

Browsing, comparing, ready to buy, asking a price, asking wholesale terms, needing help, reporting a problem. The shopper never sees the machinery — they just get an answer that fits what they asked.

A sales team at a shared table, several on headsets

A daily summary, not a ticket queue.

What shoppers asked, which products drew the most questions, which questions it could not answer, and who is waiting for a reply — information most small sellers never collect at all.

Conversations312
Ready to buy47
Escalation rate9%
Median response4s

Numbers you can actually argue with.

Conversations, ready-to-buy shoppers, conversion where your platform allows it, escalation rate, response time, usage. Pilots that fail usually fail because nobody could measure them. These are the measures.

Safety & handoff

The best thing it does is stop.

You sell things that plug into a wall and get hot. Recalls, injuries, fires, burns, electrical faults, anything about use around small children — these are never answered from general knowledge and never guessed at. They go straight to a person, and the shopper is told so. It is a default, not a setting you have to find.

Try to break it
Data & control

Your catalog, your customers, your data.

The assistant handles your product and pricing data and personal details volunteered by shoppers. That makes data handling part of the product, not a page in the contract.

Only what the sale needs.

Contact details only when a shopper wants a follow-up, a quote or a notification. It never asks for card numbers or government identifiers, and it tells shoppers not to send them.

Encrypted, and separated per seller.

Your catalog, conversations and leads are kept apart from every other seller's. Access is limited to your own users and the few Tora Labs staff needed to run the service.

Yours to export or delete.

You own the catalog, the conversations and the leads. They are kept on a defined period written into the terms, exported whenever you want them, and deleted on request. Tora Labs does not sell customer or visitor data.

Sub-processors on the record.

The AI providers and hosting we rely on are listed and reviewed, so swapping a model provider never quietly changes where your data goes.

Straight answers

The questions people actually open with.

Short versions here. The agent will take any of them further, and will tell you when it does not know.

Do you have customers yet?

No. Hiulix is in internal testing and Tora Labs has no paying customers and no revenue. The founder is checking answer accuracy against realistic appliance questions, the website embed, the lead logic and the analytics before taking money for it. What is open right now is a free pilot of up to 60 days, run by the founder personally — five to eight stores to begin with.

Why only appliance sellers?

An earlier version of this company built a general website sales chatbot for technology and professional-services firms. That market is crowded with well-funded vendors and a $50,000 company has no obvious way to stand out in it. Small appliances are different: product data has a predictable shape, the questions that decide a sale repeat, the seller base is fragmented and growing, and nobody in it can win on price. A product built for one trade can be useful on the first day instead of after a configuration project.

How is this different from Tidio, Gorgias or Intercom?

Those are established products with large customer bases, and they are a reasonable choice. They are general e-commerce chat and help-desk tools, so the appliance-specific part — capacity and wattage and which filter fits which model, fit checks before purchase, dealer enquiries — is something you configure yourself. Hiulix ships with that structure and a sales console instead of a ticket queue. That is a positioning claim, not a proven advantage: a larger competitor could add appliance templates in a few weeks, and the company says so in its own plan.

What stops it inventing a specification?

It answers product questions only from your catalog, manuals and policies. When a question falls outside them the designed behaviour is to say so and offer a handoff, not to improvise. It also makes no safety or regulatory claim — that something is UL listed, say — unless you supplied it as data. This is the most important constraint in the product: a confident wrong answer about wattage or fit costs you a return, and a wrong answer about a heating element could cost more than that.

What if a shopper reports a burn, a spark or asks about a recall?

It stops. Questions about recalls, injuries, fires, burns, electrical faults or use around infants and children are never answered from general knowledge. The shopper is told a person will respond and the conversation goes to your designated contact immediately. You are asked for that contact and your recall information at onboarding, because this is the one part of the product that cannot be configured later.

How long does setup take, and do we need a particular platform?

No particular platform. At launch you install a lightweight website component and upload your catalog as a spreadsheet or CSV along with your policy and FAQ documents, so it works for any seller with a website. Live catalog and stock through Shopify and WooCommerce connectors are planned for roughly months 7 to 12.

Is voice available?

No. A voice add-on is planned for Year 2 — the obvious uses are the shop phone and the dealer line — but it is not built and the site will not claim it is. Voice costs more per interaction, is less forgiving of a wrong answer, and raises its own consent and call-recording questions, so text comes first.

Will it connect to Shopify or WooCommerce?

That is the plan for roughly months 7 to 12, for live products, stock and order status. Nothing is live today. Ask the assistant about a connector by name and it will tell you that honestly rather than nodding along — which is the behaviour worth testing before you trust it with a shopper.

Who is behind this?

Tora Labs LLC, a Pennsylvania company (entity 0014963635, filed October 30, 2025), founded and wholly owned by Nodirbek Anvarov, who capitalised it with $50,000 of his own money and built the MVP himself. There are no outside investors. More on the company.

Built for one trade. Start with a free pilot.

Bring a real question from your own store to the assistant and see how it handles it. If it goes well, the next conversation is with Nodirbek.

Talk to the agent

No card, no form, no sales sequence. It is a conversation.