Back to Blog
Perspective

The marketplace model: an open agent ecosystem for FMCG

FMCG Cloud Team · Research8 min read

FMCG software has spent two decades arriving one point solution at a time. A distributor buys a field-sales app from one vendor, a route-planning tool from another, a shelf-recognition product from a third, and a business-intelligence stack from a fourth. Each comes with its own contract, its own login, its own data model and its own definition of what a "store" or an "order" even is. The result is a route to market stitched together at the seams, where the hardest engineering problem any operator faces is not running the business but reconciling the systems that are supposed to describe it.

We think that pattern is the thing to break. The Industry Cloud for FMCG is built on a single conviction: the entire consumer-goods route to market deserves to run on one data model, with software that behaves like a connected ecosystem rather than a drawer full of disconnected licences. That is what the marketplace model is for. First-party agents and vetted partner solutions live in the same place, on the same data layer, under one contract, one login and one bill. The structure is the product as much as any individual feature is.

Start with why "one data model" is not a slogan. Underneath every category we offer — Field Sales, Retail Execution, B2B Ordering, Route and Delivery, Revenue Growth AI and Shelf Intelligence — sits the ConnectX data layer. When a rep logs a visit, when a shelf photo is scored, when an order is placed or a route is sequenced, every event reads from and writes to the same canonical shape. A promotion an agent recommends in Revenue Growth AI is the same promotion a field rep sees in their morning brief, because there is only one record of it. Integration stops being a project you fund every year and becomes a property of the platform you bought.

VendorsEnterprise buyersVendorVendorVendorVendorVendorFMCG Cloudone platformBuyerBuyerBuyerone contract · one login · one bill

That shared foundation is what makes a genuine marketplace possible rather than just a directory of logos. A directory points you at vendors and leaves you to do the wiring. A marketplace where everything sits on one data model means a partner's solution is interoperable on arrival. A trade-finance product can read invoice and stock signals without a bespoke pipeline. A boardroom dashboard can render on the live model with no separate modelling layer to maintain. A specialist promotion or forecasting agent can act on the same facts the rest of the platform already trusts. The customer is not buying integration risk alongside the software; the platform has absorbed it.

We are agents-first about this on purpose. The primary unit on the marketplace shelf is a working AI agent that does a job end to end — audits the shelf and triggers the reorder, plans the territory and sequences the route, scores the promotion before it runs and loses money. Classic apps, data products and integrations are listed alongside, and they matter, but the centre of gravity is the agent that takes an action rather than the report that describes one. This is where FMCG Cloud Intelligence comes in: the brand intelligence layer that lets these agents reason over the shared model and hand work to one another instead of operating as silos that happen to share a login.

An open ecosystem only earns trust if openness does not mean a free-for-all. That is the job of two structures we treat as foundational. The first is the FMCG Cloud Agent Taxonomy: sixteen agent types organised into five families. Every solution on the marketplace classifies somewhere in it, which means a buyer can reason about what a listing does and how it behaves before installing it, the way a well-organised category in any store tells you what to expect from a shelf. The second is FMCG Verified, the certification every solution must earn. Verification covers protocol compliance on the shared model, observability with an auditable decision trail, performance against per-family expectations, a security review with least-privilege access, a support commitment, and a documented data-handling posture aligned with regimes like GDPR and CCPA. First-party and partner solutions clear the same bar. The mark is the reason a buyer can trust any listing, regardless of who built it.

The economics have to align with all of this, or the openness is rhetorical. A platform that taxes its partners heavily quietly pushes them to build walled gardens and route customers off-platform — which is exactly the fragmentation we set out to end. So founding partners pay a 0% platform fee. We make our return on the demand the platform generates, not by metering the rails an honest ecosystem needs to be free. For a builder, one integration puts a product in front of consumer-goods buyers with contracts, billing and AI interoperability handled. For the customer, the partner's incentive is to be excellent on the shared model, because that is what gets a solution discovered, verified and adopted.

None of this rests on claimed outcomes; the case is structural. Fragmentation is a real and well-understood tax on FMCG operators, paid in reconciliation, integration projects and decisions made on data nobody fully trusts. A marketplace on one data model, organised by a real taxonomy, gated by real certification and aligned by economics that reward openness, is the structure that removes that tax instead of repackaging it. That is the bet we are making, and we would rather be judged on the architecture than on a number we cannot yet honestly show.