Cybernetic Hive Mind

"Please track my hours for this client project." "Start a marketing campaign based on the work we've done and our social networks." At Rosenblatt, every one of us hands requests like these to a digital twin, and watches it learn the task as it goes.

The challenge

Every company loses time to human slowdowns: hand-offs, status updates, work waiting on whoever knows how to do it. The surprising part is that you don't know which parts are broken until everything goes through a twin. Then the gaps show up immediately.

What I built

Every employee has a digital twin: an agent that works on their behalf. The twins are connected into a hive mind around a shared company brain, so a skill one twin learns, or something one of them remembers, is available to the rest. Twins review their own work and update their skills as they go.

How it came together

A twin is only useful if you can trust it not to say or do what its human wouldn't. Two mechanisms handle that: evaluations that tune each agent's personality and behaviour, and human-in-the-loop approval, where a twin sends any action it's not confident about to its human first.

The team was on board from the start: a core group of engineers who believed in the approach. My security background shaped the foundations: strong identity, secrets management and network hardening, so the twins could be given real access safely.

What happened

Every digital task we do has accelerated since the twins took on the work.

What I'd take forward

A visitor-facing version, such as "meet the company brain" in an experience center, would be a published slice of the brain: reviewed before publishing, so no sensitive information leaves, while visitors still talk to a small, approved knowledge base.