Build, Buy or Both? How Agencies Run AI Agents on Client Work

At most agencies, AI lives with one person. They’ve connected the ad accounts and written the prompts, while the rest of the team still works the old way. Every time a model updates or a connection breaks, that one person has to fix it.

That’s usually when the question comes up: keep building your own, buy a tool, or run both?

In this session, two agency leaders explain how they made that call and got their whole team using AI.

What you’ll learn:

• When building your own AI setup makes sense, and when the maintenance starts costing more than it saves
• How to get the whole team working with the same agents, knowledge and connections
• How to run several AI tools side by side, and which jobs to give each one
• What to have in place before putting client data in front of an agent

Hear real stories from:

• Sjoerd van Harmelen, owner at Sababa, who built SababaHub, one of the most advanced in-house AI setups we've seen. He shares what he’d build again and what he’d never want to maintain again.
• Anna Alexiadi Smyrniotaki, Director of Digital Analytics & Strategy at GKV, whose team cut monthly reporting time by 75% with AI agents. She shows how they use AI for client work today.

Both run one of their agents live, so you’ll come away with two workflows you can adapt for your own agency.