About · Salewise
Built in London.
Close to the aisle.
Salewise builds the assistant a shopper can reach from a shelf — and the record a merchant can use afterwards.
Why we exist
The highest-intent moment is the least staffed.
Most retail software is built for head office. The shop floor gets the leftover interface: a search box, a stock lookup, a queue.
A shopper standing in front of a product with a question is the best moment a retailer will get. It is also the moment when the shop floor is busiest, the queue is longest, and the person who knows the answer is restocking another aisle.
Hardware-first answers — kiosks, smart mirrors, electronic shelf labels — add a procurement project to a labour problem. They cost thousands per unit, take months to install and need someone to maintain them.
The phone is already in the shopper’s hand. A QR code costs nothing to print. The missing piece was an agent good enough to answer from the merchant’s own catalogue, in the shopper’s language, without inventing a price. That piece is now good enough.
Salewise is the loop around that: scan, ask, decide, pay at the till — with every question recorded and every attributed sale labelled by how it was matched.
The phone assists. The till sells. The record is the product.
The Salewise position
The point of view
Five positions that shape the product.
01
Software, not hardware
The shopper already carries a capable phone. Kiosks, mounts and firmware are a procurement decision the store does not need.
02
An answer, not a search box
Shoppers describe a job to be done. The agent asks, ranks and explains — the way a good associate would.
03
Grounded, not generative
The catalogue is the source of truth. The model selects and phrases; the server renders price, stock and variants.
04
Staff amplified
Routine questions go to the agent. People take the moments that need judgement, context and a human.
05
Honest measurement
Exact matches are exact. Modelled matches are labelled as modelled. The subscription is the only fee.
London, specifically
London is the test, not the market.
The first deployments are here because the shops are here, the POS estate is fragmented, and we can sit in the store and watch the first week.
London has a dense independent retail base, a high cost of labour and a shopper who already scans. It is a hard place to build a retail product and a good place to be held to account.
The city is also where the product gets its first real test: not in a pitch, but at a shelf on a Saturday, with a queue at the till and a shopper who wants a straight answer about fit, stock or a gift under £40.
GS1 Sunrise 2027 makes a scannable 2D code mandatory on every UK pack. The QR habit is becoming infrastructure. Salewise is being built for the point where the code is ordinary and the answer is the differentiator.
The product is London-first while it is proven. That focus is deliberate: one retail culture, one POS landscape, one set of shopper expectations to get right before adding another.
What that focus buys
- Catalogue first
- Every answer is checked against the merchant’s own product record.
- One loop
- Scan, ask, compare and pay at the till — no new checkout.
- Honest labels
- Exact and modelled attribution are reported differently.
- A short rollout
- A supported provider can be live once catalogue access is granted.
How it is built
The model is rented. The loop is ours.
Salewise is a product company, not a model company. The engineering is in the catalogue, the rendering and the record around the answer.
01
Retrieval before generation
The server chooses candidate products from the merchant’s catalogue before the model is asked anything. The model can only select from that set. It never searches the open web.
02
The server owns the numbers
Price, stock, variants and offers are rendered from catalogue records. A number in the model’s prose is never the source of truth.
03
One adapter per provider
Each POS integration is a file and a registry entry with honest capability labels. The Sage UI renders from registry metadata, not a hardcoded switch.
04
The demo has to run without a database
With no database the API still runs on a seeded repository; with no model key the assistant reports that it is unavailable. Production fails closed without its required secrets.
The stack, plainly
Fastify and TypeScript on the API, PostgreSQL or Cloud Spanner behind a repository layer, React in the shopper PWA and the Sage console, and a hosted model through an OpenAI-compatible API. One adapter per provider, one product across every store.
Honest footing
Where we are. Where we are not.
A young company should be legible. This is the version we would want to read.
Where we are
- A working product, deployed with London retailers.
- Native Salewise integration; four POS adapters in beta.
- 29 languages supported; all 29 enabled in the demo.
- Founder-led sales and onboarding.
- No hardware, no shopper app and no card data.
Where we are not
- Not a certified enterprise vendor. No SOC 2 or ISO 27001 today.
- Not a payments company. No card data, no checkout.
- Not a hardware company. No kiosks, mounts or firmware.
- Not in the US yet. The product is focused on London retail today.
- Not claiming a sales uplift we cannot measure from the merchant’s own data.
Team
Founder-led, close to the customer.
The work is done by the people who will answer your email.
Akhil Raj Arimbra
Founder, CEO & CTO
Today at Salewise. Owns product, engineering, the retailer relationship and the commercial model — and built Salewise's first AI MVP.
Before Salewise. 7+ years building enterprise solutions for SME retailers and hospitality at Zensign Global. BSc Computer Science. Scalable SaaS and applied AI.
Nicholas Anderson
Chief Operating Officer
Today at Salewise. Oversees operations and partnerships, making sure every deployment delivers measurable value in store.
Before Salewise. 16 years in B2B communications and partnerships at E Seek Inc. (CA), working with business customers and channel partners. BA, UNLV.
For company information, email support@salewise.co.uk.
If you run a shop
Start with one aisle.
Create an account, connect the catalogue, and see the first questions your shoppers ask. If you would rather talk it through first, the team is on the contact page.