eSentriX is a no-code platform that automates the turnup of smart e-commerce agents — with built-in cross-agent support (ChatGPT, Gemini, Grok) and LLM-powered personalization and targeting that boost sales and turn your agentic integration into a smart sales powerhouse.
eSentriX understands shoppers in their own words — multilingual, human-language intent that finds the right product in any market. We transform raw retail catalogs into LLM-ready intelligence, orchestrate agent-to-agent (A2A) commerce from discovery to fulfillment, and let retailers automate merchandising and optimization inside their own environments — no rebuild, no rip-and-replace.
This is how AI becomes commercially reliable.
In production with
Haat DeliveryKing StoreHomeAweBetakBringy
orders placed through the agent
of conversations end in an order
of orders take the agent's upsell
bigger basket when they do
01 · Where it sells
Wherever the shopper is, the same agent is already there.
One catalog, one brain, one set of rules — surfaced on your site, inside your app, on WhatsApp and on the phone. Scroll through the six places it sells.
yourstore.com
Website
One script tag. Your site starts selling.
The assistant floats over every page and sells from your live catalog — text or voice, Arabic, Hebrew or English.
It finds across aisles, explains its pick, adds the upsell and hands off to your checkout. Shadow-DOM isolated, so your CSS never fights it.
The assistant screen ships from our servers and updates without an app release.
It never touches your cart. It emits add_to_cart, remove_from_cart and checkout; your app runs them with the code it already has. iOS, Android, React Native, Flutter.
Your WhatsApp Business number answers in the shopper's own dialect — search, cart and a checkout link.
Connect the number in one click with Meta's own flow; click-to-WhatsApp ads open the agent on the exact product the ad showed, and every order is attributed to the campaign.
A number shoppers dial. The agent talks, builds the basket and confirms — no app, no screen.
Caller-id sign-in means the shopper who ordered in the app is recognised on the phone, and "the same as last week" just works. Point your own Twilio number at us or ask for one.
A page of its own. Put it anywhere your shoppers are.
No website needed: the agent lives on its own link, in your colours, selling from your live catalog.
Drop the link in a bio, a story, a QR on the shelf, a receipt, a WhatsApp broadcast or an ad. The shopper lands in a conversation, not on a landing page — and leaves with an order.
Instead of a landing page, the ad opens a conversation that starts on the product the ad showed.
Click-to-WhatsApp and Instagram ads need nothing but your connected number — the shopper stays in WhatsApp and the agent opens selling that product. Link ads land on your agent page, open on the product too. eSentriX sees the ad click and the in-chat order, so every shekel of spend is attributed.
One request becomes a full shoppable routine — matched to live catalog items, bought in one tap, checked out, and upsold. This is the flow eSentriX runs behind any agent.
demo 01 · where the shopper meets the agent
A sponsored post that sells on the spot.
The search box that answers back.
The agent on its own — anywhere your shoppers are.
Watch: the shopper taps the ad and is inside the retailer's agent — no install, no store visit — with a real cart and an order before they scroll on.
Watch: the same agent inside the store's own search bar — a query becomes a compared, explained, filled cart.
Live now: the agent as a page of its own — a link you put in a bio, a QR on a shelf, an ad, a message. It replays real shopper requests until you type; try "a BBQ for 10" or "compare the olive oils".
Try: tap the post, ask for "snacks for tonight", accept the extras, place the order — the whole loop runs on a live catalog.
The console that runs the agent — a read-only tour of a live store.
Watch: the catalog coming in from a store link, the agent's voice and selling rules, the checkout flow drawn as cards, the channels (web, app, WhatsApp, phone) — and every conversation with the order it produced.
Try: open Agent to see the voice and rules, Flows for the checkout cards, Analytics for the attributed orders. Nothing you click changes the store.
Legacy backend in. Agent-ready infrastructure out.
Four stages abstract away the complexity of retail backends — turning fragmented systems into one protocol any agent can transact with.
🔌🔌🔌🔌
stage 01 · connect
System connectivity
Adapters plug into legacy ERPs, POS, and checkout stacks — no rip-and-replace, no storefront rebuild.
ERP / POSSOAP → RESTwebhooks
🧬🧬🧬🧬
stage 02 · enrich
Data enrichment
Catalogs are normalized and enriched — attributes, taxonomy, embeddings — so agents can reason over every deep-catalog SKU.
multilingual intentLLM-ready catalogembeddings
⬡⬡⬡⬡
stage 03 · protocol
UCP exposure
Everything is exposed through the Universal Commerce Protocol: discover, price, cart, checkout — uniform across retailers.
one APIA2A workflowsagent-native
🚚🚚🚚🚚
stage 04 · execute
Transaction & fulfillment
Agent intent becomes a real order: payment captured, inventory committed, last-mile dispatch triggered automatically.
real-timedelivery routing<80ms p95
04 · Personalization engine
It learns the shopper, then shops like them.
Every message teaches eSentriX something — taste, budget, who they're feeding, what they can't eat. Those signals become a living profile applied to every result: re-ranking what fits, hiding what doesn't, and flagging what's unsafe.
👤
Shopper profileinferred from chat · updates live
● learning
Interests
🔥 BBQ & grilling🥗 home cooking🏋️ fitness💄 skincare
Allergies & restrictions enforced
🥜 nut-free🥛 lactose-free
Diet
🌿 2× vegetarian✡️ kosher🍬 low-sugar
Context
👨👩👧👦 family of 4💳 ~₪250 / order⭐ premium brands
How the agent applies it — automatically
🚫
Filters the unsafe
Nut and dairy products never appear in results or carts.
↑
Surfaces what fits
Vegetarian and kosher options rank first for the household.
◎
Re-ranks to taste & budget
Picks tuned to interests and a ~₪250 basket — 97% match.
🔔
Warns across agents
If any connected agent adds an allergen, it's flagged before checkout.
05 · The moat · Signal loop
The agent doesn't just sell. It turns every conversation into growth.
Because eSentriX is the sales agent, it captures first-party intent no pixel or QR code can see — then feeds it straight back into the retailer's ads and campaigns. A loop that compounds with every message.
01👂
Listen
The agent captures intent in language — what shoppers want, reject, their budget, reorder cadence and life-stage — plus every real purchase.
→
02🎯
Target
Those signals become phone-matched audiences, synced to Meta, TikTok, Instagram and Klaviyo — high-intent, not guesswork.
→
03✨
Create
The agent generates the campaigns and creative itself — it already knows what each shopper wants and how they ask for it.
→
04📈
Measure
Closed-loop attribution: eSentriX is the only layer that sees the ad click and the in-chat purchase — proving what every shekel of spend returned.
Why it's defensible
Anyone can wire an AI to checkout. Only the agent inside the conversation owns the signal — and every cycle makes the next campaign smarter. That compounding first-party data is the moat.
06 · Universal Commerce Protocol
Integrate once. Transact everywhere.
One standard interface between any agent and every connected retailer — so nobody builds point-to-point integrations again.
⬡
One protocol, every retailer
Agents call the same five endpoints whether the backend is a modern stack or a 20-year-old ERP. eSentriX handles the translation.
⚡
Real-time, transactional
Live inventory, live pricing, atomic checkout. Agent intent moves to physical delivery without a human in the loop.
🛡️
Trust & control built in
Agent authentication, spend limits, and retailer-side policy controls — so supply partners decide exactly what agents can do.
📈
Validated in live pilots
Already proving in production that retail backends can be abstracted and made instantly compatible with AI shopping assistants.
agent.ts — UCP client
// Any agent. Any retailer. One protocol.constoffers =awaitucp.discover({intent:"weekly groceries",budget:250,deliverBy:"fri 19:00"})// → matched across connected retailersconstorder =awaitucp.checkout({cart:offers.best.toCart(),agent:credentials,limits:policy})awaitucp.fulfill(order.id)// → inventory committed · payment captured// → last-mile dispatched · ETA 18:40 ✓
07 · Pricing
Plans for growing and enterprise brands.
Every plan runs on the same core — catalog intelligence, multilingual search and the Universal Commerce Protocol.
StarterFor a single store getting started
GrowthFor growing brands
ProFor omnichannel retailers
EnterpriseFor platforms & large fleets
Commerce agents
AI shopping agent (web & text)
✓
✓
✓
✓
Voice agent
—
✓
✓
✓
WhatsApp channel
—
—
✓
✓
White-label agent (your brand, your name, your domain)
—
✓
✓
✓
Phone line (call-in ordering)
—
—
—
✓
Agents included
1
3
5
Unlimited
Catalog intelligence
Catalog ingestion & enrichment
✓
✓
✓
✓
Multilingual semantic search (AR · EN · HE)
✓
✓
✓
✓
Visual & image search
—
✓
✓
✓
Real-time inventory & pricing sync
—
—
✓
✓
Platform & integrations
Store connectors (Shopify, WooCommerce, Salla)
✓
✓
✓
✓
Embeddable storefront widget (SDK)
✓
✓
✓
✓
MCP tool layer & API access
—
✓
✓
✓
Agent-to-agent (A2A) orchestration
—
—
—
✓
Custom domain for the agent
—
—
—
✓
Personalization & analytics
Conversation analytics dashboard
✓
✓
✓
✓
Shopper profiles & recommendations
—
✓
✓
✓
Merchandising rules & selling notes
—
✓
✓
✓
Meta ads that sell on the spot (ad → agent → order)
—
—
✓
✓
Campaigns & automations
—
—
✓
✓
Scale & support
Catalog sync frequency
Daily
Hourly
Real-time
Real-time
Uptime SLA 99.9%
—
—
—
✓
Support
Email
Email
Priority email
Dedicated + phone
Pricing is tailored to your catalog size and session volume — request pricing and we'll come back within one business day.
The next web won't be browsed. It will transact.
Whether you own catalog, delivery capacity, or agents — plug into the infrastructure that connects all three.