Trade Me

Trade Me builds an AI-first ecosystem on Google Cloud to turn data into teammates

Results on Google Cloud
  • Connects directly to BigQuery to extract complex marketplace trends with natural language and decommissioned 100% of physical data centers to modernize operations on Google Cloud

  • Reduces idea-to-production lead time to half a day using a Thinnest Viable Platform (TVP) supported by Google Cloud

  • Achieved 4 9s availability for critical APIs through Cloud Run and AlloyDB

  • Slashed incident triage by up to one week with custom Gemini agents built on the Agent Development Kit (ADK)

  • Targets 20% efficiency gains through Gemini Enterprise Agent Platform and agentic coding tools

Trade Me leverages Google Cloud integrated AI platform to break data silos, empowering employees to drive digital agility and competitive differentiation.

Moving from legacy silos to a
unified engine of execution

Close-up of a blue smartphone screen showing the Trade Me and Property app icons, featuring a stylized kiwi bird logo

Trade Me, a household name as an online marketplace and classifieds platform in New Zealand, is far more than an auction site. It is a vital digital ecosystem connecting millions of Kiwis across every major life stage. From finding a first job to buying a family home, Trade Me acts as the primary data pulse across a multitude of New Zealand’s economic sectors, including property, motors, employment and consumer habits.

Historically, the organization faced a critical challenge: years of operating on monolithic on-premise infrastructure had created a foundational bottleneck. To solve this and accelerate its AI transformation journey, Trade Me launched its Google Cloud migration. This foundational step was the essential prerequisite to allow the team to eventually build platform engineering artifacts. Eventually, the team was successfully able to turn off the last piece of hardware in its data centers, completing a total exit from physical infrastructure to build long-term operational muscle. To transform its vision into a scalable reality, the organization sought an integrated platform capable of uniting its data, security, and infrastructure to deliver trusted, agentic AI at scale. By engineering a high-performance modular architecture using Cloud Run and AlloyDB, Trade Me ensured the shift was not merely about moving workloads; it was about resilience and building the stage for the next wave of innovation.

We decided to lead our cloud transformation internally because we are a technology company at heart. This internal leadership allowed us to drive a migration where critical systems were rebuilt to better leverage the capabilities of Google Cloud. This also ensured our people gained the skills needed to effectively operate in the cloud during the transition.

Paolo Ragone

Executive CTO, Trade Me

Scaling execution with a custom agent ecosystem
on Google Cloud

A laptop shows the Trade Me website with marketplace, property, and car listings against a vibrant purple background.

To drive competitive differentiation and eliminate delivery bottlenecks, Trade Me initiated the development of its Thinnest Viable Platform (TVP). Following a platform-as-a-product mindset, this concept was designed to serve as a ‘golden path’ to production-ready applications.

By abstracting away the underlying complexity of infrastructure, TVP actively reduces cognitive load for developers. This empowered engineers to quickly build new capabilities on a secure, modular foundation, bypassing the constraints of legacy systems to slash the idea-to-production lead time to just half a day. Building on this foundation, the organization later turned its focus toward generative AI. Moving beyond standard no-code tools, the team leveraged the Agent Development Kit (ADK)—experimenting with the library before its public release—to build a fleet of custom agents deeply integrated with internal APIs.

This agentic architecture utilized Gemini Enterprise Agent Platform (formerly Vertex AI) and Gemini Enterprise app to perform the heavy lifting across the business. By wrapping existing APIs in the ADK, Trade Me deployed specialized agents tailored to departmental needs, including an Incident Agent designed to retrieve and compare log information across multiple systems. This streamlined the triage process, saving a week or more of diagnosis time and allowing the team to focus on high-value development. Simultaneously, the organization democratized insights by integrating agents with BigQuery, allowing employees to ask complex marketplace questions in plain English.

The significant unlock has come from the combination of the entire Google Cloud ecosystem. By using Gemini agents to pull in and assess data from observability tools, we have saved a week or more of triage and diagnosis time in some cases. From an operational perspective, that speed is gold; from a security perspective, it is platinum.

Kieran Cornelius

Head of Platforms, Trade Me

Redefining the marketplace
through velocity and vision

A person in a bed holds a smartphone showing a real estate app, with a breakfast tray blurred in the background.

The cloud migration, the shift to the TVP architecture, and the adoption of agentic AI have cumulatively redefined the heartbeat of how Trade Me operates. The organization has unlocked an unprecedented level of speed, slashing the lead time from an initial idea to a production deployment to half a day.

And this new velocity ensures that Trade Me can react to the market in real-time, backed by an operational reliability that has seen critical APIs move from 3 9s to a robust 4 9s of availability. This technical foundation is now driving a target of 20% efficiency gains across the engineering organization through the use of agentic coding tools and Gemini Enterprise Agent Platform.

As an impact of this multi-year journey, Trade Me has undergone a profound human transformation. With 60% of internal testers already reclaiming 30 minutes of their workday, the workforce is shedding the burden of routine ‘toil’ and shifting toward higher-value strategic roles. This democratization of data is further realized through agents that connect directly to BigQuery, allowing employees across the business to extract complex marketplace trends using natural language rather than specialized code. 68% of users report that these tools have made their jobs more enjoyable and fulfilling.

Looking ahead, Trade Me is already charting its next course: evolving from a traditional search-based marketplace into a proactive personal assistant that anticipates a consumer’s needs across their entire life journey. By combining Gemini, Google Cloud, and Google Workspace into their ecosystem, the organization is building a definitive roadmap for a highly autonomous and intelligent future in the digital marketplace industry.

We are proud of the excitement we’ve managed to boost across the organization. We see our teams deconstructing their tasks to delegate cumbersome work to AI, which represents a permanent shift in how they think. Google Cloud removes the burden of data privacy and security risks, allowing us to focus entirely on solving actual business problems.

Nicolò Musmeci

Head of AI Engineering, Trade Me

Founded in 1999, Trade Me is New Zealand’s leading online marketplace and classifieds platform. It connects more than five million members across major life stages, including property, motors, and jobs.

Industry: Retail and Consumer

Location: New Zealand

Products: Gemini Enterprise Agent Platform, Gemini Enterprise app, BigQuery, Cloud Run, AlloyDB, Cloud Code, Agent Development Kit (ADK), Google Workspace

Google Cloud