Skip to content

Leadership1 publisher3 min readPublished

AI agents are giving rank-and-file workers a manager's job at their old pay

Staff are doing manager work over AI agents without manager pay or title, Entrepreneur reports, with each engineer at startup Daytona overseeing about five. Companies are counting the output gains, and leaders now have to decide whether supervising agents counts toward job level and pay.

The Board Room · Leadership desk

Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction

Illustration accompanying AI agents are giving rank-and-file workers a manager's job at their old pay
Generated illustration

What happened

  • Business Insider's reporting has workers telling agents what to do, reviewing their output, asking for changes and stepping in when something goes wrong.
  • Daytona chief executive Ivan Burazin said the startup's engineers no longer write code themselves.
  • Wipro's chief technology officer, Sandhya Arun, pictures a future in which one engineer manages a group of AI agents.
  • Salesforce has introduced 'job-ready agents' for sales, customer service, ecommerce and internal operations, along with tools for employees to train them.

Compiled by The Board RoomSomething wrong?How this is made

Why it matters

  • cost If job levels stay as they are, the employer keeps the value of the extra output while the staff supervising the agents carry the coordination load at their existing pay.
  • exposure Staff who check AI-generated work stay accountable for everything published or sold, so a reviewer takes the consequences of errors in output they did not write.
  • precedent Vendors are shipping agent-training tools for ordinary employees to use, so supervising agents becomes part of standard jobs and more employers will have to settle the leveling question.

Daytona's chief executive, Ivan Burazin, told Business Insider that his engineers oversee about five agents each on average. Across 16 engineers, that comes to roughly 80 agents [8][1]. The company has 30 people in all [8].

Sinda Khenine, a software and AI engineer at Electrolux, described what that oversight involves [11]. She picks the models, gives them context and rules, checks their work and assembles the output of several agents into a finished product [11]. "The effort is more focused on the coordination itself rather than the engineering problem that we are solving in the first place," she told Business Insider [13]. Khenine also manages people at Electrolux. She said overseeing agents can be even more demanding, because they need extensive setup and continued attention once they start working [12]. "It's like a loop," she said. "You need to monitor the results, you need to see the progress. You need to decide if you need to rerun or finish the job." [14]

The strongest objection to the unpaid-manager reading comes from Khenine herself. She said agents helped her write more code and build a company on her own, something she said would have been "10 times harder" before [10]. That gain is clearest in the AI business she is starting for herself, where the extra output belongs to her [10]. Inside an employer, the firm counts the output. Wipro's chief technology officer, Sandhya Arun, told Business Insider that AI has added the productivity equivalent of 20,000 workers [6].

The pay side of the argument has thinner evidence behind it. Entrepreneur's summary says the new duties come without a promotion or a raise [1]. Business Insider reports that employees are under pressure to use the tools and now spend less time on the work of their professions [2]. The reporting does not include salary figures, leveling data or attrition rates. Named people describe the change in duties first-hand. The pay gap is a general characterisation, and nothing in the record links it to people quitting. I think this belongs in this year's leveling cycle. The evidence does not support treating it as a retention emergency this week.

For leaders, the trade-off is payroll cost against a job description that matches the work. Whatever goes into this year's job descriptions decides what next year's performance reviews measure. Counting agent supervision as management raises the pay base. It also means drawing a line on when overseeing about five agents, as Daytona's engineers do, makes someone a manager [8]. Leaving it out keeps costs flat. The reviews then go on scoring hands-on craft, the part of the job Khenine now spends less time on [11].

What to watch

  • Pay or title data from any employer showing whether agent supervision is being written into job levels or compensation.
  • Attrition or survey figures that link agent-oversight workloads to people leaving; the current reporting has none.
  • Whether employers using Salesforce's agent-training tools assign that work to formal, paid roles or spread it across existing staff.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories