₹30cr seed bet that became a $40M Series B in 5 years

In 2021, a Bengaluru startup raised ₹30 crore in seed funding to bet that the internet would eventually move past passive video. Five years later, that same company closed a significant Series B led by QED Investors — with Shah Rukh Khan joining the round. The journey from AR social app to AI interactive content platform is a masterclass in staying obsessed with a problem while being ruthless about the product form.

The Pivot Nobody Talks About

The ₹30 crore seed round in July 2021 — led by Anand Chandrasekaran, 9unicorns, and Inventus Capital Partners — was for an AR-based social networking app. Think Snapchat filters meets Indian social. The team was small, revenue was early-stage, and the product was consumer-facing.

By 2026, none of that consumer framing survived. What did survive was the core obsession: content on the internet doesn't have to be a one-way, passive experience. Flam's founders — Shourya Agarwal, Amit Gaiki, and Malhar Patil — killed the social app and rebuilt around enterprise brands that need interactive, personalised content at scale.

This is the pivot most coverage skips. It wasn't a technology pivot. The underlying belief — that interactive beats passive — never changed. What changed was who they were selling to and what problem they were solving for that buyer. Consumer social is a distribution game; enterprise content is a conversion game. Flam chose the latter.

For any founder watching this: the seed round bought them the time to figure out the right customer, not just the right technology. That's what early capital is actually for.

Three Formats, One Thesis

Flam's product line is three formats — Flicks, Airboards, and Visual Agents — and each one attacks a different place where static or passive content loses a buyer.

Flicks lets a brand shoot one video and serve multiple variations of it — swapping the character, product, or scene — with fast switching latency. For a D2C brand running ads across Hindi, Tamil, and English markets, that means one production budget, not three. AI compression cuts asset size by 60%, which directly reduces CDN costs.

Airboards streams high-fidelity 3D content through a camera interface — no app download required — with a 300ms time-to-first-buffer. A furniture brand can let a customer point their phone at a wall and see a sofa in their actual room, without asking them to install anything. The no-app-download constraint is a deliberate business decision: every friction point between a consumer and a product experience is a drop-off.

Visual Agents creates a hyperreal avatar from a single image, capable of speaking in multiple languages with natural expressions. A bank's customer service, a brand's spokesperson, a retailer's in-store guide — all running from one image, no studio required.

The thesis connecting all three: every piece of content on the internet should be interactive. That's not a feature. That's a category claim.

Five Models, Own the Margin

Most AI startups in India are resellers. They wrap GPT-4 or Claude in a UI, charge a margin, and call it a product. The risk is obvious: the moment OpenAI ships the feature natively, the wrapper disappears.

Flam built five proprietary models instead: Fable 2.0 for product animations, Fantom 1.0 for identity-preserving video synthesis, Falcon 1.0 (a large language model), Finesse 1.0 for multilingual speech synthesis, and Forge 1.0 for video editing and object swapping.

The business implication is straightforward. When you own the model, you own the cost structure. You're not paying per API call to a third party every time a brand runs a campaign. At enterprise scale — thousands of interactive ads, millions of impressions — that difference compounds into real margin. It also means Flam can price on value (what the brand earns from a higher-converting interactive ad) rather than on cost-plus (what the API call cost them).

Building five models before raising a large round is the kind of capital discipline that makes a Series B conversation very different. You're not asking investors to fund the build — you're asking them to fund the scale.

This mirrors what we've seen with lean, focused teams elsewhere: the Monk CI team raised ₹4 crore to own their CI infrastructure rather than depend on a third-party pipeline — same logic, different category.

Latency as a Business Moat

The technical numbers Flam publishes matter only insofar as they change buyer behaviour. Here's why they do.

A consumer will abandon a mobile experience that takes more than 3 seconds to load. Interactive 3D content that buffers for 2 seconds feels broken. Airboards' 300ms buffer means it feels like video — the experience is seamless enough that a consumer doesn't notice the technology. That's the bar. Not "fast for 3D" but "indistinguishable from a video scroll."

For enterprise buyers — CMOs at FMCG, retail, and auto brands — this translates to a simple question: will this work in a live campaign without embarrassing us? Fast character swapping in Flicks means a consumer can switch between product variants in real time without a visible load. That's the difference between a gimmick and a tool a media buyer will actually put budget behind.

Flam was at Cannes Lions 2026, where Jitendra from Google and Tariq Hassan (former McDonald's CMO) both spoke about the platform publicly. That's not a startup showing up at a conference — that's a product that enterprise buyers are already using and willing to endorse on the record.

What SRK Actually Signals

Shah Rukh Khan joining the Series B gets the headline, but the real signal is distribution, not celebrity association.

Flam's formats — especially Visual Agents — are built for exactly the use case where a brand needs a recognisable face speaking to consumers in multiple languages, at scale, without a production shoot every time. SRK as an investor is also SRK as a potential use case, and as a door-opener to every major FMCG, telecom, and entertainment brand in India that already works with him commercially.

This is a pattern worth noting for founders: the right strategic investor doesn't just bring capital, they bring the first ten enterprise conversations you couldn't cold-email your way into. For a platform selling to CMOs, a CMO-adjacent investor with existing brand relationships is worth more than the cheque.

The same logic applies to QED Investors leading the round — they're a fintech-focused fund, which signals Flam is already in conversations with banks and financial services brands for Visual Agents deployments. Customer service, financial product explainers, multilingual branch communication — all of these are active use cases for a hyperreal avatar that speaks multiple languages.

Staying in the Problem

Five years is a long time to stay committed to a thesis that the market hasn't fully validated yet. Most founders either give up or drift — adding features to chase adjacent revenue, raising a bridge to survive, or pivoting so hard that the original insight is lost.

Flam's arc from ₹30 crore seed to a major Series B is notable precisely because the thesis didn't change. The internet should be interactive, not passive. What changed was the go-to-market: from consumer social (where distribution is everything and margins are thin) to enterprise content (where conversion is everything and margins are defensible).

The early team stayed obsessed with the same problem long enough to build five proprietary models, prove the formats with real enterprise brands, and show up at Cannes Lions with Google and ex-McDonald's CMOs talking about the product on camera.

For founders building in AI right now, this is the uncomfortable truth: the companies that will matter in five years are the ones that stayed in their problem long enough to own something proprietary — a model, a dataset, a distribution channel — rather than the ones that moved fastest to wrap the latest API. As we've seen with organic-led growth outperforming cold outreach for funded startups, the compounding advantage almost always goes to the team that builds something hard to replicate.

Enterprise is the Real Buyer

Flam's Series B use of capital is explicit: R&D, product expansion, and scaling global enterprise sales. Not consumer acquisition. Not app downloads. Enterprise sales.

The Cannes Lions 2026 presence is the clearest signal of who Flam is selling to. Paul Woolmington, CEO of Canvas Worldwide (one of the largest independent media agencies in the US), appeared on Flam's own podcast. Tariq Hassan, who ran marketing at McDonald's, did the same. These aren't vanity conversations — they're the people who sign the contracts.

For an Indian AI startup, winning enterprise CMO attention globally before the Series B closes is a meaningful proof point. It means the sales motion is working, the product is enterprise-ready, and the category is real enough that buyers are willing to be publicly associated with it.

The Series B will fund the sales team and the infrastructure to serve those enterprise contracts at scale. That's a very different risk profile than a consumer app burning cash on user acquisition — and it's why QED Investors, a fund that understands unit economics deeply, led the round. This is also why early-stage startups that nail their positioning before scaling tend to convert enterprise conversations faster — the product story does the selling.

What Founders Can Steal

Flam's journey has three lessons that apply well beyond interactive content:

1. Own the infrastructure before you scale the sales. Five proprietary models before a major Series B means every enterprise contract is margin-accretive, not margin-dilutive. If your unit economics depend on a third-party API you don't control, you don't have a business — you have a distribution channel for someone else's product.

2. The right pivot preserves the thesis, changes the customer. Flam didn't pivot away from interactive content — they pivoted away from consumer social to enterprise. The problem stayed the same; the buyer changed. That's a go-to-market pivot, not a product pivot, and it's far less destructive to team morale and product quality.

3. Strategic investors are distribution, not just capital. SRK and QED both bring customer conversations Flam couldn't reach otherwise. When you're choosing your Series B lead, ask: who does this investor talk to every week that I need to talk to?

The One Question

Every founder building in AI content right now should ask: if the underlying model I depend on ships this feature natively next quarter, does my business still exist?

Flam's answer is yes — because the value isn't the model alone, it's the format (Flicks, Airboards, Visual Agents), the distribution infrastructure (web embed, shareable links, Flam codes for physical channels), and the enterprise relationships already in place. OpenAI shipping better video generation doesn't make Flam's switching latency, 3D streaming, or multilingual avatar infrastructure irrelevant. It might make the underlying models cheaper to run.

That's the difference between building a product and building a wrapper. Five years and a major Series B later, Flam is clearly the former.

Conclusion

Flam's story is simple: stay in the problem long enough to own something proprietary, then sell to the buyer who has the budget to pay for it. The ₹30 crore seed bought five years of model-building and market-finding. The Series B funds the scale. If you're building in AI right now, the question isn't whether your product is impressive — it's whether you own the infrastructure underneath it.