Federico Jan

Bio

Most brands rent attention: they pay for it with ads, borrow it from algorithms, and lose it when the budget stops. I'm interested in the other kind, the attention that shows up when someone searches Google, asks ChatGPT, or looks for a how-to on YouTube. I call the work of earning it Attention Ops, write about it at We Dream Labs, practise it with others in the Attention Ops Lab, and build it into Gomnee, which helps experts turn what they know into why customers find them.

Before that, I spent two years as the first engineer, and for most of that time, the only engineer, on the agent side of Marketing.MBA. I joined during the Claude 3.5 Sonnet era, before there was an industry standard for any of this, and owned the architecture of REMI, a production multiagent GTM operating system, end to end. By the time I moved on from full time work there, the team had grown to four engineers.

The common thread was production engineering: tracing thousands of agent runs, supporting workflows that stayed stable for more than an hour, working across more than five agent frameworks, and building the CLIs, knowledge systems, approval gates, and integrations around them. Context engineering mattered, but it was one part of making the whole operating system reliable.

I also worked directly with employees to bring agents into daily workflows. That forward deployed work meant handling the hard conversations around trust, adoption, ownership, and business reality, not only the technical architecture.

That work is also where Attention Ops started. One of the systems I built there was an agent native SEO department that produced more than 40 in depth articles with human approval, an early version of the loop Gomnee runs today.