unthinkable
Retail & commerce tech · AI Product Engineering

Help brands and retailers get every shopper from looking to buying, without wondering.

For companies building software for brands and retailers

We help you build the product that makes it possible: AI that helps shoppers find the right item, answers the questions that stop a purchase, and keeps them informed until it arrives, with stock and prices read live by code.

The retail problems a product can take on

52%of shoppers left a store without everything they came for: sales a cross-store stock search can saveZebra, 2025
20%of US shoppers who abandon a cart blame slow delivery, a doubt a delivery date at checkout can answerBaymard Institute, 2025
71%of US shoppers are less likely to shop again after a poor return: the moment an exchange flow can win backNRF and Happy Returns, 2025
What we build

What we build for retail and commerce technology.

The commerce products we build again and again. For each: how it helps shoppers find what they want and buy it, what AI does (and what people still decide), and how we help you build it.

Store & omnichannel

Shoppers find the item in any store or online, get help at the shelf to choose it, and know when what they bought will arrive.

BUILD AI IN

An assistant finds stock across stores and holds or ships it, answers at the shelf, and gives staff each customer's history and preferences so they can help them choose. Staff decide what to offer.

BUILD WITH AI

Connected to POS, stock and loyalty systems, so store and online read the same records.

Shipped forClientela, PDI Technologies

Retail analytics & trade promotions

Teams see what made shoppers buy, and which promotion worked and why, before the next one is planned.

BUILD AI IN

Questions about what sold, what ran out and which promotions moved shoppers to buy, answered in plain words with the numbers behind each answer; unusual results flagged.

BUILD WITH AI

Figures computed from your data by code, never guessed by a model; dashboards on the warehouse you already run.

Shipped forGMG, PDI Technologies

Brand storefronts

Shoppers find the right product, get their doubts answered before checkout, and buy it.

BUILD AI IN

Product questions answered from your catalogue, search by photo or look, the delivery date and return terms shown before checkout, and a wrong size swapped in conversation; your team decides refunds and exceptions.

BUILD WITH AI

One storefront joining stock, loyalty, stores and fulfilment, built so checkout holds up on sale days.

Shipped forShoppers Stop, The Souled Store, Urban Ladder, Foodstories, InTouch, Ziora Beauty

Running a brand or a retail chain? Brands & retail →

Your customers' journey

The same shopper journey, as your product sees it.

Six moments where shoppers stop to wonder before they buy. At each: what your product does, what it has to connect to, and what a retail buyer will check before signing.

Today: a wait before every stepWith AI at work: straight through

1 · Finding it

Your productSearches stock in every store and the warehouse, then places a hold or a ship-to-home order from the same chat.

Connects toPOS and order-management APIs (Shopify POS, for example), with the store's hold and transfer rules.

Retail buyers checkThat counts and hold limits come from the POS by code, and a stale count is flagged rather than promised.

Store & omnichannel ↓

2 · Choosing in store

Your productAnswers a shelf-code scan with stock, colour and care details, and pages an associate when the shopper wants one.

Connects toThe product catalogue, per-store stock and the associates' handhelds.

Retail buyers checkThat a handoff to a person is one tap away, and product details come only from the catalogue.

Store & omnichannel ↓

3 · Buying online

Your productWorks out a delivery date from stock location and carrier transit times, offers ship-from-store when that is faster, and states the return terms beside the buy button.

Connects toThe storefront and checkout (Shopify Storefront API or the platform's own), warehouse stock and carrier transit data.

Retail buyers checkThat every date and term shown traces to stock, carrier or policy data, never to the model.

Brand storefronts ↓

4 · Where is my order

Your productWatches tracking events, turns a delay into a plain-language message to the shopper, and opens a ticket with the order history when a case needs a person.

Connects toUPS, FedEx and USPS tracking, the order system, and help desks like Gorgias or Zendesk.

Retail buyers checkThat the expected date always comes from tracking data, and each escalation reaches a person with the full order history.

Store & omnichannel ↓

5 · Delivery day

Your productLists open delivery windows, rebooks on request and sends a text before the crew arrives.

Connects toThe delivery-scheduling or last-mile partner's booking system.

Retail buyers checkThat windows are read from routing and capacity rules, so the product never offers a slot the crew can't make.

Store & omnichannel ↓

6 · Swapping it

Your productLooks up the order, applies the return-window rule, offers an exchange before a refund, and books the label or store drop-off.

Connects toOrder and returns systems, warehouse receiving and the payment provider (Stripe, for example).

Retail buyers checkThat refunds above a set limit wait for a person, and card numbers never pass through the agent.

Brand storefronts ↓

What the product does at each moment (text version)

Finding it

  • Understand the product (Name, photo or link) → Check stock everywhere (Stores and warehouse) → Hold or ship (POS and order system) → Store sets it aside (Hold confirmed). Guardrail: If stock data is older than the store's last count, the agent says so and asks the store to confirm.

Choosing in store

  • Shopper scans a code (At the shelf) → Answer the question (Stock, colors, care) → Check stock and route (POS and who's on shift) → Associate helps in person (Fitting, styling, complaints). Guardrail: Complaints, damaged items and anything the product data can't answer go to a person.

Buying online

  • Understand the question (Dates, returns, fit) → Work out the real date (Stock location and transit times) → Read the return terms (Written policy) → Terms set by your team (Fees, windows, exceptions). Guardrail: The agent never promises a date or a term the systems don't show.

Where is my order

  • Understand the question (Any channel, any wording) → Read tracking (Order system and carrier APIs) → Explain in plain words (What happened and what's next) → Service team handles claims (Lost parcels and compensation). Guardrail: If a parcel has had no scan for 3 days, the agent opens a case for the service team instead of guessing.

Delivery day

  • Order ready to deliver (From the warehouse) → Talk to the shopper (Windows, access, questions) → Book the window (Route and crew capacity) → Delivery lead confirms routes (Changes on the day). Guardrail: Access problems and special handling go to the delivery lead before the day.

Swapping it

  • Understand the request (Size, color, reason) → Check the policy (Window, condition, final sale) → Reserve the replacement (Live stock) → Exceptions go to a person (Outside policy or damaged). Guardrail: Anything outside policy goes to customer service; the agent never makes exceptions.

Try it with your own case →

Why products stall

Why commerce products stall, and how we help you avoid it.

BUILD AI IN
“It will promise stock we don't have.”
Agents read live data through tools; prices, discounts, stock and refunds are computed in code, so the model cannot promise what the store cannot deliver.
“We can't tell if it's helping.”
We test agents on real product questions and edge cases from your catalogue, and track answer accuracy and how many shoppers go on to buy, before a wider launch.
“It will replace our support team.”
It takes the repetitive questions, such as stock, sizing, order status and returns within policy, so your team spends its time helping the shoppers who need judgement.
BUILD WITH AI
“It won't fit our platform.”
We build on Shopify's Admin and Storefront APIs, app extensions and webhooks, and on marketplace and ERP APIs, respecting rate limits and app review rules.
“It will cost too much to run at scale.”
We cache, batch catalogue jobs and use smaller models where they are enough, and report model cost per conversation and per merchant.
“It won't be ready before the season.”
A senior lead and a complete team from week one, on building blocks already proven in production, so a first release ships.
You inherit

Building blocks already built.

Already built
Shopify app scaffold with billing and webhooksin production
Catalogue search and retrieval layerin production
Policy engine for refunds, discounts and price limitsin production
Merchant approval queuein production
Conversation and action login production
Cost-per-action meteringin production
Your retail & commerce tech productstarts on these, not from scratch

The team around it

Product designerShapes how people and agents work together
AI engineerAgents, prompts, evaluations
ArchitectRecords, integrations, security
DevelopersBuild it with AI, keep it maintainable
TestersSpecs and tests that prove it works
Domain expertKnows your industry's rules and edge cases
Security reviews

Will your product pass a retailer's security review?

Brands and retail chains review a vendor's data handling before any pilot. These are the answers we build in from the first release.

BUILD AI IN
Answers a reviewer can trace
Every price, stock level, date and refund the agent gives is read through a tool or computed in code, and logged with its source.
Tested against prompt injection
Shopper messages are tested for prompt injection before every release, so a chat cannot change a price or a policy.
People on the exceptions
Refunds above the merchant's limit, complaints and price changes go to a person with the order and its history.
BUILD WITH AI
Card data stays with the payment provider
The agent never sees or stores card numbers; card handling stays with the provider (PCI DSS).
Shopper data on the shopper's terms
Read with the signed-in shopper's or merchant's permissions and never used to train models, under GDPR and each market's privacy rules.
A test store before live orders
Built and tested in a test store or staging environment first, with the retailer's security review planned from the start.
Integrations

The integrations retail buyers ask for.

A brand or retail chain will ask what your product connects to before it asks about features. These are the connections we build.

Shopify and commerce platforms

Shopify Admin and Storefront APIs, app extensions, webhooks and Shopify POS, within rate limits and app review rules; marketplace APIs where merchants sell through them.

Stores, stock and ERP

POS, stock and loyalty systems, and ERPs such as SAP, so store and online read the same records.

Carriers and delivery

UPS, FedEx and USPS tracking, plus last-mile booking systems.

Service desks and payments

Gorgias or Zendesk for handoffs with the order history, and payment providers such as Stripe for refunds.

You own the result: code, prompts, evaluation sets, data and the app listing sit in your accounts, so your product's IP stays yours.

Proof

Commerce products we have shipped, and what they did.

A US fashion e-commerce company15,000+products sold through image search (as reported)

An AI shopping assistant: shop the look from a photo, virtual try-on and outfit suggestions. Shoppers couldn't find clothes that matched a look they liked, and left their carts.

Read the case study →
A leading Indian online beauty retailer40%add-to-bag rate from shoppable videos (as reported)

AI-tagged shoppable videos, so shoppers add the product straight from the video, plus a rebuilt blog and a community app. Shoppers watched its videos but had to go and search for what they saw.

Read the case study →
V2 Retail, India (300+ stores)300+stores fulfilling online orders (as reported)

Omnichannel commerce that uses every store as a fulfilment point: live stock sync, an order dashboard and a store app for scan-checked picking. It had no online store, so online shoppers couldn't find or buy what its stores had.

Read the case study →
Shoppers Stop, India100%uptime during peak traffic (as reported)

Rebuilt web and mobile storefronts and a new CMS in four months, without replacing its commerce stack. On peak sale days, slow pages and broken checkouts stopped shoppers buying.

Read the case study →

Results as stated in each linked case study; figures marked "as reported" to be checked against the source before launch.

Which part of the buying journey would your product fix first? Name the step where your shoppers drop off: search, product pages, checkout or delivery. After a short discovery you get a plan with dates for the first release, measured against that drop-off.

Tap one: Ask Unthinkable starts your plan from there. Or say it in your own words ↓

How to choose

How to choose who builds retail & commerce tech products.

Four things a commerce founder or CTO should check, and how we meet them.

Has shipped commerce products

Storefronts, store and omnichannel products and retail analytics, for companies such as Clientela, PDI Technologies and GMG, and for brands like Shoppers Stop.

Often elsewhereAgencies that ship themes and apps, with little AI depth.

Knows the retailer's side

We also help brands and retail chains run their stores and service day to day, so we know the workflows your merchants expect a product to fit.

Often elsewhereEngineers who have never watched a store team at work.

AI that acts inside the product

In a store product, for example: stock found across stores and held or shipped from one chat, with staff deciding what to offer.

Often elsewhereA chat widget bolted onto the storefront.

Ready for peak season

We set the architecture before the prototype and keep AI writing inside it; prices, discounts and refunds stay in code, and model cost is reported per conversation and per merchant.

Often elsewhereA demo that slows down or gets costly once real catalogues and sale-day traffic arrive.

The full comparison: us, other software companies, in-house teams and consultants →

Ways to work

Start the way that suits you.

Time & materials

A senior AI team billed for time, within an agreed monthly budget. Even on time and materials the work is tied to your business objective: every plan starts from it, and every report shows the number next to the hours.

One fixed outcome

One agreed outcome at a fixed price, with acceptance criteria, including the measure, written up front.

Dedicated team

A complete senior AI team, monthly, working as part of yours and reviewed each quarter on what shipped and what moved.

Linked to the result

Where the gain can be measured from your own systems: a reduced implementation fee plus a percentage of the expected gain, for a set period.

How we work and price →

Straight answers

Questions retail and commerce teams ask.

How do I choose a development partner for a commerce product?

Test them on what your merchants will test you on. Ask how their assistant gets a price or a stock level: the answer should be a live tool call, with discounts and refunds computed in code, never the model's memory. Ask how they evaluate it on real catalogue questions and edge cases before launch, and how they keep model cost per conversation low enough for each merchant to stay profitable for you. Ask what they have built on Shopify's Admin and Storefront APIs, webhooks and app review, or on marketplace and ERP APIs, and whether store stock and fulfilment are in scope or only the storefront. Finally, check that the code, the app listing and the data sit in your accounts. Unthinkable has built storefronts for Shoppers Stop, store-led omnichannel for V2 Retail's 300+ stores and an AI shopping assistant for a US fashion company, and it also helps brands and retailers run, so it knows what your merchants will ask.

How do I add an AI shopping assistant to my Shopify store or app?

Start with the questions that stop shoppers buying today. We connect an agent to your catalogue, stock and order data through Shopify's APIs, keep prices and policies in code, and test it on real questions before launch. For app builders, we package it as a Shopify app with billing and extensions so merchants can install it.

Can an AI agent give wrong prices or promise stock we do not have?

Not if it is built correctly. The agent never states a price or stock level from memory; it calls tools that read live data, and discounts and refunds are calculated by code within limits you set. If the data is missing, it says so and hands off. We test these cases specifically, because a confident wrong answer about price costs more than no answer.

How much does it cost to build a Shopify app with AI?

It depends mostly on how many systems the app connects to and how much setup each merchant needs, not on the AI model. A first release is usually one fixed outcome, and you get a written estimate after a short discovery. Unthinkable measures running cost per conversation during that first release, so you can price the app and each merchant stays profitable for you. You own the code and the app listing.

Can you build seller tools for a marketplace?

Yes, so shoppers can find and trust what sellers list. Common starting points are listing creation, catalogue cleanup, onboarding checks and pricing suggestions. The agent drafts; marketplace rules run as code; the seller approves. We log every change so your trust and safety team can see what the agent did. We integrate with your marketplace backend and with sellers' own systems where they connect.

Will AI agents replace my customer support team?

They take the repetitive questions, such as stock, sizing, order status and returns within policy, so your team spends time helping the shoppers who need judgement. Every conversation can be reviewed, and the agent hands over with a summary when it is unsure or the customer asks for a person. Most teams use the time saved for better service rather than fewer people.

Can you add AI to our existing commerce platform without rebuilding it?

Yes. Unthinkable adds agents and AI features through your platform's existing APIs, catalogue and order data, so the storefront, checkout and back office keep working as they do. We start in a test store or staging environment, ship one feature that answers a question shoppers or merchants ask today, and measure it before adding the next. A rebuild is worth it only when the platform itself blocks what you need, and we say so plainly.

How do you keep shopper and payment data secure?

Card data stays with your payment provider, so the agent never sees or stores card numbers. Shopper data is read with the signed-in shopper's or merchant's permissions, never used to train models, and every action is logged. We design for GDPR and the privacy rules of each market you sell in, and test for prompt injection before every release.

Ask Unthinkable

Where do shoppers give up in your product?

Tell us about your product and where shoppers stop before buying. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with a senior AI engineer.

Send us a note instead

Required

Thanks. A senior AI engineer will reply to you.
u
Ask UnthinkableClarity first, then the right team
Your conversation

An AI agent. This prototype uses prepared answers and sends nothing.