When we began our AI journey, we explored which use cases could deliver meaningful day-to-day value. Six months later, the picture is far clearer. While some early experiments didn’t stick, three AI solutions have emerged as potential game-changers for our business:
- Lead enrichment
- Publisher health checks
- Inbox management
At the heart of each of these is Google’s Gemini Enterprise, the enterprise platform that now powers our agentic AI systems and saves us hours every week. So let’s dig in.
Lead Enrichment: How One Reimagined Workflow Sparked Company-Level Efficiency
Sales is where Gemini Enterprise looks the most “agentic” for us right now, because we’ve paired it with an ADK-built agent (Agent Development Kit) that does the legwork a human usually gets stuck with. In plain terms, it pulls data from multiple sources and enriches publisher leads with external context, so the business development team can spend less time on manual enrichment and more time moving through their pipeline.
The best part is how it came to life. One of our BD reps learned Python, figured out what an agile workflow needed, and became one of our advanced AI champions. With the right support behind her, that effort turned into something genuinely practical: lead enrichment that used to take up to two hours can now be done in minutes.
And because we’ve already built the workflow and tooling into that ADK agent, we can repurpose it in other teams, too. With inventory quality back in focus throughout 2025, our Audience team has been able to repurpose that lead agent to help identify inventory quality based on the data the agent can connect to, and the tools it has to work with said data. It’s a nice example of how one workflow can start in Sales, then turn into shared infrastructure for other teams without starting from scratch.
Publisher Health Checks, Supercharged
Our Publisher Support Management (PSM) team sits on the front line. They’re often the first stop when something feels “off” on a site, and they need to move quickly without missing anything. Being a more technical team, they were able to take an ADK template and build their own agent, without needing as much support to build it out.
They use it to support our existing weekly routine health checks on a random set of sites to confirm everything is set up optimally. The agent helps perform some of those checks, reducing cognitive load and speeding the team up, with a knock-on effect for publishers as we work to keep setups optimized.
We’re also being deliberate about keeping humans in the loop. Some checks are still verified by a person while we build trust and expand capabilities. When we run the agent from Gemini Enterprise, it has memory, so the context windows get saved. This means that when we run into issues, we can reopen the thread and continue troubleshooting at any point. Long-term, this is the kind of foundation we want for more sophisticated agents in 2026.
Inbox Management: The Digital Assistant Revolution
The third solution tackles something we all feel from time to time: email overload. Gemini Enterprise has native integration with Google Workspace, enabling sophisticated inbox management that goes far beyond basic filtering. It analyzes messages, interprets sentiment, and generates prioritized action lists tailored to each role. For instance, a customer success manager’s inbox looks very different from an engineer’s, so now we have a system that can be configured accordingly – all through no-code agents.
For me this has been great. Instead of my usual routine of checking my email and calendar first, I’m able to go straight to Gemini Enterprise and ask what my day looks like. It tells me if a colleague has changed a meeting time, so I can attend on time, rather than catching it too late in a cluttered inbox or calendar.
Preparing for AI-Driven Change
These three solutions represent far more than efficiency gains. They signal a fundamental shift in how work gets done.
We’ve established an AI ethics committee, bringing together executives, legal, people operations, and technical leaders to navigate this transition thoughtfully. The committee addresses critical questions such as the use of ‘vibe coding’, AI-assisted development by non-technical team members. We’re seeing fantastic innovation from people who aren’t traditional engineers. They’re building working solutions to problems they face daily.
Naturally, we’re working hard to close the gap between a working proof of concept and a production-ready system with proper development protocol, but we’re getting there. What’s been key for me is that although we’re sometimes quick to lose trust when AI gets something wrong, it’s important to remember all the times it gets things right and the time that saves us.
We’re on the cusp of a major shift in how we work. So far, we have achieved a lot without incurring high costs, and we are already seeing significant value from these basic implementations. As we continue testing Gemini Enterprise’s value proposition, the early results suggest a future where AI augments human capability rather than replacing it. The goal isn’t fewer people; it’s people freed from repetitive tasks to focus on strategic, creative work.
In Q1, we expect to have much clearer insights into the platform’s long-term potential for transforming how we operate – and I can’t wait!
