Valentina Akpan: Founder, Rellatech. Admin and technical virtual assistant for businesses and startups with teams.
The New York Times ran a piece in June, picked up by Entrepreneur, on how small business owners are shifting from using AI as a tool to managing it as a workforce. AI agents now handle client intake, routine inquiries, and document organisation. In one example, a bankruptcy lawyer described his agent stack completing hours of admin work in minutes but operating faster than he could track, and his job, he said, had become supervising outcomes and reviewing decisions made by systems he did not fully control.
I read that and thought: yes, that is exactly the problem.
Agents do the work, but supervision is a different job entirely.
An agent that runs client intake is great until the day it tags 40 leads as the wrong customer type and none of them get a follow-up, and an agent that organises documents is great until it starts filing every new contract under "Other" with no alert because a template changed. These failures do not throw errors, they just happen, and you find them three weeks later when a client emails to ask why nobody got back to them.
Supervision is a role rather than a tool or a dashboard, and it needs someone who understands the systems, checks the outputs on a regular cadence, catches drift, and takes responsibility for the outcome when something goes wrong.
The bankruptcy lawyer problem
The owner in the NYT piece said the plain part out loud, which is that the system was faster than he could track, and that is not a personal failure but what happens when the operator layer is missing. The founder ends up being the one supervising, which means they are back to the exact problem the AI was supposed to solve, and it just looks different.
This is where an operator earns their keep. As a CRM and automation specialist, my job on an AI-heavy account is to run the audits, watch for unnoticed failures, tune the prompts and workflows when the business changes, and take responsibility for the outcome so the founder does not have to.
If you have agents, hire the supervisor
You do not need a big team to run an AI workforce, and you need one senior person on a retainer whose job is to keep it running cleanly. That is the version of technical operations that holds up over 12 months, and the alternative is what the NYT quoted, which is a founder becoming full-time overseer of software they do not fully understand.
Work with Rellatech
If you're running agents and workflows already and worried about what you aren't seeing, that supervisory work is a regular part of what I do. See how the plans are structured, or a short call is the fastest way to see whether it's a fit.
Contact MeNot ready for a call? Take the free operations assessment and I will send you a written breakdown within 24 hours.