Breaking
SandStar AI Retail - the store that sees Cashierless shopping at ~5% the cost of Amazon Go 99%+ product-recognition accuracy, no RFID or barcodes Deployed across roughly 30 countries Powers Coca-Cola's Smart Lounge in Beijing WEF Technology Pioneer, 2021 Founded 2017 as YI Tunnel, rebranded SandStar in 2020 Backed by True Digital Group / C.P. Group
Company Profile · AI Retail

SandStar AI Retail

The store that sees. Computer vision that recognizes what you pick up - and bills you when you walk out.

Founded 2017 Charlotte, NC · Beijing Series B+ ~60 employees Computer Vision
SandStar AI Retail company logo
SANDSTAR AI RETAIL - The brand rose "from sand and dust, reaching for the stars." Led by founder-CEO Yili Wu, a Tsinghua software graduate and ex-Oracle, ex-IBM engineer.
0
Countries served
99%+
Recognition accuracy
0
Fortune 500 clients
4
Core product lines
The Dispatch

A camera, a neural net, and no checkout line

SandStar AI Retail builds the machinery of the store that runs itself. Where a shopper once queued to scan barcodes, SandStar puts cameras and deep-learning models: the system watches a customer enter, tracks what they lift from the shelf, notices what they put back, and generates the bill after they leave. There are no RFID tags, no weight sensors, no handheld scanners. The company describes the approach in plain terms - retail seen through the "eyes" of AI and run by the "brain" of big data.

The firm was founded in 2017 in Beijing under the name YI Tunnel and rebranded to SandStar in 2020, alongside a Series B+ round led by Thailand's True Digital Group, an arm of the C.P. Group conglomerate. The new name, founder Yili Wu has said, was meant to evoke rising "from sand and dust, reaching for the stars" - a signal of global ambition for a company that had until then operated largely in the Chinese market.

What sets the work apart is the insistence on pure vision. Many autonomous-store efforts lean on a mix of sensors - RFID chips on every item, scales under every shelf. Those add cost and break. SandStar's bet was to strip them out and lean entirely on computer vision paired with what it calls semantic action recognition, so the system can read the gesture of taking an item or returning it, even when several people shop at once. The company reports recognition accuracy above 99 percent.

That focus has commercial consequences. In its YI Tunnel days the company claimed it could run a cashierless store at roughly 5 percent of the cost of Amazon Go - a striking figure that framed autonomous retail as something ordinary operators, not just tech giants, could afford. It is the kind of claim that gets a startup noticed, and it did: the World Economic Forum named SandStar a Technology Pioneer in 2021.

Today the company positions itself around full-stack AI - computer vision, generative AI and agentic AI - with solutions it says are deployed across roughly 30 countries, serving dozens of leading operators and more than two dozen Fortune 500 firms. Its US operations run out of Charlotte, North Carolina, a long way from the Beijing lab where the first pure-vision store went live.

SandStar drives retail evolution through the eyes of AI and the brain of big data.
- SandStar company positioning
How It Works

Grab. Go. Done.

Inside a SandStar unmanned store, four things happen without a cashier - and without the shopper doing anything unusual.

1

See

Ceiling cameras identify each shopper's position and trajectory the moment they enter.

2

Understand

Action-semantic models read the gesture of taking or returning an item - even with multiple shoppers.

3

Tally

A virtual cart updates in real time at 99%+ accuracy, no barcode scan or RFID tag needed.

4

Bill

The customer walks out; the system closes the cart and triggers payment automatically.

SandStar states it tracks shopping behavior rather than biometric identity, and cites ISO 27001 certification and GDPR-aligned practices for its data handling.

Products & Services

Four ways to make a store autonomous

2018

AI Vending Machine

Camera-based vending using convolutional neural networks and dynamic pure-vision recognition to detect items taken and returned - marketed with a zero product-recognition-fee model and retrofit kits for existing machines.

2018

Pure Vision Unmanned Store

Fully cashierless shops that identify a shopper's identity and trajectory by camera, then auto-generate the bill on exit. No RFID, barcodes or weight sensors.

2019

AI Check-out / Smart Kiosk

Vision-based self-checkout that recognizes multiple products at once without scanning - built for staffed retail, micro markets and food service.

2020

Smart Store Analytics

A computer-vision analytics layer: customer dwell time, heat maps, product-interaction data, inventory optimization and staffing or promotion recommendations.

Business Model & Funding

Hardware, software, and a Thai telecom's bet

SandStar sells B2B: AI-enabled retail hardware - vending machines, kiosks and unmanned-store camera systems - bundled with a cloud analytics platform, plus retrofit kits that add vision to existing infrastructure. It leans on a "zero recognition fee" pitch and long-term warranty and support to win operators. Public funding data points to backing from Baidu Ventures and Kinzon Capital early on, growth capital from Guopeng, Fengshion and Mobai in 2019, and a 2020 Series B+ from True Digital Group.

Funding journey

SELECT ROUNDS · SOURCES: PR NEWSWIRE, FINSMES, CRUNCHBASE
Early / Angel
2017–18
Growth (2019)
2019
Series B+
Aug 2020

Bars indicate relative round progression, not disclosed dollar amounts. Several amounts were undisclosed; reported cumulative funding ranges from ~$15M to ~$27M across sources.

The Difference

Pure vision vs. the sensor stack

Autonomous retail is a crowded field - Amazon's Just Walk Out, Standard AI, Trigo, AiFi, Grabango, Zippin. SandStar's argument is architectural.

The SandStar approach

  • Cameras and deep learning only - no RFID, no shelf scales
  • Semantic action recognition handles multiple simultaneous shoppers
  • Retrofit kits add vision to existing vending and store hardware
  • Cited claim: cashierless store at ~5% the cost of Amazon Go
  • Behavior tracking, not biometric identity; ISO 27001

The sensor-heavy alternative

  • RFID tags on every SKU raise per-item cost
  • Weight sensors under every shelf add hardware and maintenance
  • Purpose-built stores are expensive to construct and scale
  • Higher capital outlay narrows the field to large players
  • Often tied to a single retailer's ecosystem
Customers & Partnerships

A Coca-Cola lounge that learns

In 2022 SandStar supplied the computer-vision and deep-learning layer behind the Coca-Cola Smart Lounge in Beijing - a recreational retail space where shoppers move between beverage, coffee and interactive zones while the system quietly collects behavioral data across each section. It was a marquee proof that big brands would trust vision-based analytics inside a flagship experience.

Beyond Coca-Cola, SandStar has worked with Chinese operators including the retail chain Yonghui and carrier Hainan Airlines, and reports deployments serving retail operators, brand owners, hospitality and food-service businesses and enterprises across roughly 30 countries. Its investor True Digital Group doubles as a strategic route into Southeast Asian markets.

The name begins from sand and dust, reaching for the stars - reflecting our ambition for the global market.
- Yili Wu, on the YI Tunnel to SandStar rebrand
Milestones

From Beijing lab to 30 countries

2017

YI Tunnel is founded

Yili Wu launches the company in Beijing to digitize offline retail with computer vision and big data.

2018

First pure-vision unattended store

A camera-only unmanned store goes into operation; the AI vending machine enters commercial use.

2019

Growth funding

Raises hundreds of millions of yuan from investors including Guopeng, Fengshion and Mobai as AI+Retail gains traction.

2020

Series B+ and rebrand to SandStar

True Digital Group backs the round; YI Tunnel becomes SandStar to signal global reach.

2021

WEF Technology Pioneer

Named a World Economic Forum Technology Pioneer; showcases solutions at the NAMA Show.

2022

Coca-Cola Smart Lounge

Powers Coca-Cola's interactive Smart Lounge in Beijing with its computer-vision platform.

2025

Full-stack AI positioning

Expands messaging to computer vision, generative AI and agentic AI across ~30 countries.

Watch & Explore

Demos, interviews & coverage

Product demos and founder interviews are indexed on the company's channels and press pages. Explore the primary sources below.

Frequently Asked

Questions people ask

What does SandStar AI Retail do?

It builds computer-vision retail systems - AI vending machines, cashierless unmanned stores and store analytics - that recognize what shoppers pick up and automate checkout without RFID, barcodes or sensors.

Was SandStar known by another name?

Yes. It was founded in 2017 as YI Tunnel and rebranded to SandStar in 2020 alongside its Series B+ funding round.

Who founded SandStar and who leads it?

Yili Wu, a Tsinghua University software-school graduate and former Oracle and IBM engineer, is the founder and CEO.

Who are SandStar's customers and partners?

Retail operators, brands and enterprises across roughly 30 countries, including 25+ Fortune 500 firms; notable partners include Coca-Cola, Yonghui, Hainan Airlines and investor True Digital Group.

How is SandStar different from Amazon Go?

SandStar uses a pure computer-vision approach with no RFID or weight sensors, and has claimed it can run a cashierless store at roughly 5% the cost of Amazon Go.

◆ ◆ ◆
Connect

Find SandStar