AI for retail stores,
your data stays in-store.

Footfall counting runs computer vision on your existing store cameras: the model detects each person entering and moving through the floor and counts them in real time, separating staff and repeat passes so the number reflects real visitors. Because it works on the video stream rather than a single doorway beam, it also measures dwell and checkout queue length by zone. Nodeflux uses a NIST-tested vision engine, and accuracy holds up across the lighting and crowd conditions of a working store. You get live counts you can staff and merchandise against, not an end-of-day estimate.

People counting and heatmaps do not need facial recognition: the system counts and tracks shoppers as anonymous shapes moving through the store, so it knows how many people and where they go, not who they are. That keeps it privacy-first and a better fit for retail data protection rules, because you are measuring behaviour, not matching faces against a database. Everything can run on-premise, so the video stays inside your stores. Face matching is a separate, opt-in capability you would only enable for a specific use case, never a default for counting shoppers.

Yes. The vision engine is built to run on-premise on hardware in your own store or data centre, so camera feeds and the analytics derived from them stay on your infrastructure and your customer and sales data stays in your stores. There is no requirement to stream raw video to an outside cloud to get footfall, queue, heatmap, or loss-prevention results. That on-premise model is also how Nodeflux deploys for regulated customers across 8 countries. You keep full ownership of the footage and the insights.

AI loss prevention watches for the behaviours tied to shrinkage, such as suspicious activity at the checkout and items leaving the shelf without a scan, and flags them live so a team member can step in during the moment rather than discovering the loss at the next stock-take. Catching events in real time protects margin directly and shortens the gap between an incident and a response. It leads with privacy: the system reads behaviour and stays in-store, so you are not face-matching every shopper to fight shrinkage. Over time the flagged events also show you where on the floor and in the day shrinkage concentrates.

A store operations AI assistant connects to your sales, inventory, footfall, and policy data and answers questions in plain language: what is selling, what is out of stock and why, how conversion tracked against footfall, or what the policy is on a return or a promotion. It can also turn the day's numbers into a short action list so every store runs on the same playbook before opening. Athena is the auditable assistant behind this, and every answer is sourced back to the underlying record so a manager can trust and verify it. The result is fewer manual reports and faster, more consistent store decisions.

Nodeflux is an Indonesian deep-tech AI company founded in 2016, with a vision engine ranked in the top 25 worldwide on the U.S. NIST Face Recognition Vendor Test, which makes it a strong fit for retailers, malls, and stores in Indonesia and across Southeast Asia. The platform pairs on-premise computer vision for footfall, queues, heatmaps, and loss prevention with an auditable AI assistant for sales and store operations. Running on-premise means your customer and sales data stays in your stores, which suits local data protection expectations. It is deployed across 8 countries, so the same stack scales from a single store to a national chain.