← Back to Work Barona: Scaling the AI marketing OS

Barona built their own AI marketing OS, and we gave it the structure to scale

"We had a working AI operating system for marketing context about Barona and our services and customers, but it was an initial build. We wanted a second set of expert eyes on it, to make the system compounding and give it more lasting structure. That's something that even the more advanced teams are still often missing."

Many marketing teams are still figuring out where to start with AI. Under CMO Joni Helminen, Barona’s marketing team has already made AI part of daily work. Joni is one of the clearest voices on AI in marketing.

By the time OstraconAI got involved, Barona was already well down the road. Joni had a clear idea. The team’s knowledge about the brand, the services, and the audiences should live inside the AI where anyone could reach it, not buried in scattered slide decks. He’d built a working first version himself. We came in to give it lasting structure and make it compound.

A team that knew what it wanted

Barona is one of the biggest staffing and recruitment companies in Finland, part of the Bravedo group. Its marketing covers six service categories inside four business areas, from staffing and recruitment to learning, wellbeing, and outsourcing. Barona also runs sub-brands like Saranen, Momentous, and Plugo, each a Part of Barona brand with its own distinct voice. Joni’s team of 18 writes for all of it.

Joni didn’t need a pitch on why AI matters. He knew exactly what he wanted to build.

Turning a head start into a system

Barona had already done the groundwork, and then some. The team had written down what makes Barona Barona, down to a deep library of 104 published articles and 70 customer references. They’d built their own AI skills for writing posts, putting decks on Barona’s official template, and drafting briefs, and wired it all into Claude. They already had a working marketing OS. What it didn’t have yet was the upkeep-structure for the system.

That’s where we came in. A pile of documents goes stale the moment the brand moves, and a system only works if the whole team runs the same one. Barona already had much of the material sitting in GitHub. What it lacked was structure. So we helped reorganize it into a more consistent shape an AI can read reliably, kept it in the repository the team already shared, and packaged the lot as a Claude plugin. A marketer installs it once and syncs updates with one command.

The Barona Marcoms OS GitHub repository showing the knowledge base, skills, MCP server, and setup folders
The Barona Marcoms OS repo: structured knowledge and skills the team installs as a Claude plugin.

The centerpiece is a context-update skill. Updating an AI’s knowledge normally means careful formatting that nobody has time for, so it never happens and the context rots. Here you hand the skill raw material, a sales call transcript, a new strategy deck, last week’s press release, and it reshapes that into the structured form the system expects, files it in the right place, and logs the change in version history. No developer, no formatting rules to remember. A sales call becomes a go-to-market playbook.

Results

Now anyone on the team can open Claude and get work that already sounds like Barona, with no briefing on the company first. A new hire can write a campaign brief on day one.

And the team owns the whole thing. We ran Claude Code training so they can change the knowledge base and build new skills themselves, instead of calling an agency every time something moves. And it keeps growing. New skills, live data, other teams across Barona: each is a smaller build than the last. The hard part, teaching an AI who Barona is, is already done.

GET IN TOUCH

Want to accelerate what you've already started?