Google Cloud AI Offerings
A unified catalog platform built for Google Cloud Next 2026 event, consolidating TCS's AI offerings into a single, searchable surface so that the right solution could be found before the conference badge had time to cool.
What It Is and Why It Exists
Large organisations accumulate AI solutions the way Victorian naturalists accumulated specimens: prolifically, enthusiastically, and with a near-total indifference to whether anyone else can find them afterwards. TCS had built a considerable number of AI offerings on top of Google Cloud, distributed across teams, verticals, and internal repositories, each one perfectly useful in isolation and collectively forming something closer to an archaeological site than a product catalog.
The platform built for Google Cloud Next was an attempt to remedy this. A unified catalog: one place to browse, filter, and request demos of AI offerings spanning the breadth of what TCS had built on Google Cloud, designed to be ready in time for one of the year's more consequential cloud events.
The Interesting Parts
The first problem was schema. Offerings arriving from different teams arrive with different shapes different levels of documentation, different conceptions of what a "use case" means, different assumptions about what a prospective customer needs to know. Normalising these into a coherent catalog without flattening the distinctions that make each offering legible required some care. The FastAPI backend, backed by Google Cloud Storage, handles ingestion and serves a consistent interface over what is, underneath, a heterogeneous collection.
The second problem was discovery. A catalog that requires you to already know what you're looking for is not much of a catalog. Search and filtering by use case and industry meant that a visitor arriving with a vague problem rather than a specific product name could still navigate toward something relevant which is, in practice, how most people at conferences actually arrive.
The frontend sits in Next.js, built for an environment where first impressions are expensive and attention is short. The interface needed to feel considered rather than assembled, which at a live event is less a vanity and more a precondition for anyone taking the thing seriously.
Tech Stack
Next.js and TypeScript on the frontend. Python and FastAPI on the backend, with Google Cloud Storage handling the catalog data layer.