Field notes on governed, agentic AI.
Original writing from the team building Humael — what we are learning shipping real, governed AI across software, voice, telecom, finance and energy.
Agentic AI vs. automation: why human-in-command wins in the enterprise
Automation repeats what you already know how to do. Agentic AI decides. Here is why the enterprises that win pair autonomy with a human in command.
From 12,000 alarms to 31 incidents: rethinking the NOC with agentic AIOps
Most AIOps tools just turn down the volume on alarm noise. Closing the loop — detect, correlate, explain, act — is a different game.
The 2% problem: the customer insight hiding in every call
Most contact centres analyse a couple of percent of their calls. The other 98% is where the churn signal lives.
Cloud or on-premise? How to choose where your enterprise AI runs
SaaS is the fast path. But for regulated, sovereign or air-gapped data, on-prem is not a compromise — it is the requirement. Here is how to decide.
The cheapest money you are not collecting: continuous three-way match
The revenue leaking out of your own contracts and invoices is the lowest-risk margin in the business. Most teams catch it weeks late, by hand.
Make energy a real-time decision: NILM, V2G and demand response
One of your biggest controllable costs is nearly invisible until the bill lands. It does not have to be.
See the products behind the writing.
Every post is grounded in a product that is live today. Take any one for a walkthrough on your own data.