Flam.

The scene, today

A magazine page begins to move.

A reader opens a print magazine in a dentist's waiting room. There is a Mahindra ad on page nineteen - the kind of full-bleed photograph that has existed, more or less unchanged, since 1968. They lift their phone, scan the small square in the corner, and the SUV on the page lifts itself off the paper. It rotates. It opens its doors. A configurator slides in. No download. No app store detour. No friction. Just paper, then physics.

That moment - the small, slightly improbable moment when ink becomes interaction - is what Flam sells. The company has done it more than 580 million times now, for over 100 of the largest brands on earth, and it has done it without ever asking a single consumer to install anything.

The problem they saw

Advertising had a tap-out rate of one hundred percent.

By 2021, the math of digital advertising had become unkind. A YouTube pre-roll was something you measured by how fast a finger could find the corner. A banner was something a brain learned to delete on reflex. Television, against all odds, was still where billions of dollars went to die quietly.

The augmented reality industry, meanwhile, had spent a decade promising relief and delivering Snapchat dog ears. AR was either a parlour trick or a 3GB Unity download, and consumers - politely, almost apologetically - had decided they wanted neither.

Three friends at BITS Pilani in India looked at this and asked an obvious question that nobody seemed willing to answer out loud. if AR is so magical why was every demo gated behind an app nobody would ever install?

The founders' bet

Three undergraduates, one stubborn idea.

Shourya Agarwal, Malhar Patil and Amit Gaiki met as students. Their bet was specific enough to be testable and large enough to be interesting: if mixed reality could be made to run inside a phone's browser, at consumer-grade frame rates, on the back of a QR scan, then every poster, package, billboard and television frame in the world quietly became programmable.

It is, on inspection, a delightfully unfashionable bet. Most metaverse companies of the era were busy building goggles for futures that never arrived. Flam went the other way - it built for the device already in the user's hand, and assumed the future would mostly look like the present, only slightly weirder.

The product

An AI-native publishing layer pretending to be a marketing tool.

From the outside, Flam looks like adtech. From the inside, it is a stack of generative AI infrastructure with a marketing org bolted to the front. The platform sits on three proprietary model families - each named, sweetly, like a forgotten band from the 1970s.

The customer-facing product is bundled into three verbs: Engage (the brand campaign side), Connect (visual agents for communication) and Convert (immersive storefronts where the 3D thing you tapped becomes the thing you buy). The plumbing underneath is the interesting part - Kubernetes, WebGL, ARCore, iOS SDK, OpenGL, PyTorch, JAX, Apache Beam, Weights & Biases. It is the kind of stack that makes infrastructure people nod approvingly and finance people quietly check the AWS bill.

The proof

Numbers that survive a skeptic.

Most AR companies have demos. Flam has receipts. The customer roster reads less like a startup deck and more like a duty-free terminal - Google, Samsung, Emirates, Mahindra, Wargaming. The kind of brands that do not pay for whimsy.

A chart that is, mercifully, mostly bars going right. Bars going right are how a Series A becomes a Series B.

The mission

Make the world tappable.

The official Flam mission is a perfectly reasonable corporate sentence about AI-native content and life-like immersive visuals. The actual mission, the one you can derive from looking at how the company spends its money, is a little blunter. Flam wants every flat surface a brand has ever printed on - paper, screen, packaging, billboard, t-shirt, boarding pass - to be a launch button for a three-dimensional thing.

This is not, despite appearances, a metaverse pitch. There is no headset. There is no parallel world. There is only the world you already live in, with slightly more stuff in it, summoned on demand. The company is careful to call it mixed reality, not virtual reality - the difference being that one cuts you off from your environment, and the other layers a digital monkey onto your kitchen counter when you scan a cereal box.

Why it matters tomorrow

The bet on the next $5B of ad spend.

Inc42 has called Flam's market the $5 billion mixed reality ad opportunity. Whether that number proves correct is, charitably, anybody's guess. What is harder to argue with is the direction. Brands are out of attention. Consumers are out of patience. The most likely place new advertising spend goes is wherever someone has solved the install-friction problem - and Flam has, demonstrably, solved it.

The risks are real and worth naming. Apple and Google could decide tomorrow that WebAR should be slower. A larger model lab could ship a comparable rendering stack. The macroeconomy could decide that experimental ad budgets are not, in fact, necessary. Flam's defense is the boring kind - 100 paying enterprises, real ARR, a working sales motion and a stack other companies have not bothered to build.

Which brings us back to the dentist's waiting room.

Back to the scene

The page is still moving.

T he reader closes the magazine. The SUV folds itself politely back into ink. They have not downloaded anything. They have not given anyone an email address. They have, however, just experienced an advertisement they will mention to somebody later today, which is a sentence almost nobody has written about advertising since the late nineteen-nineties.

Flam did that. Quietly, at scale, in the browser, with three named AI models and a team of about a hundred and fifty people split between San Francisco and Bengaluru. The skip button is still there. It just has slightly less to do.

Watch / Listen

Interviews & product demos.