🔍 Read the full analysis: Why I Ask AI Agents For Decisions Rather Than Status on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Thorsten Meyer’s October 27 article argues that AI agents can create more work when they generate updates people must read and manage. Citing Anthropic’s study of its own workforce, Meyer proposes four practices: account for human follow-up, experiment without committing to every idea, set stopping rules for recurring work, and limit the decisions a coordinating agent brings forward.
In an October 27 article, Thorsten Meyer argues that people using AI agents should ask them to surface decisions, not send routine status updates, because reading and managing agent output can consume the time the tools are meant to save. His recommendations draw on an Anthropic study of its own workforce, which Meyer says found that employee-directed agents spent about 80 percent of session time waiting for human input.
Meyer’s central point is that generating output may be cheap while reviewing it is not. An agent can produce a report or proposal quickly, but a person still has to read, check and act on the result. If several agents send updates that each seem useful, the combined workload can become a separate job.
He proposes four operating rules. First, count the human work that follows an agent’s output: identify who will review it, who will act on it and whether the resulting work is wanted. Second, use agents for small, low-cost experiments, but do not treat every promising test as a commitment. Third, set an end date and a “nothing new” stopping condition before making a task recurring, and pause it if nobody reads the results.
His fourth rule is to give a coordinating agent an attention budget. Meyer recommends silence as the default and no more than three decisions a day, each presented with a recommendation, supporting evidence and the consequence of taking no action. That shifts agent communications from a record of activity toward a short list of matters requiring human judgment.
The new bottleneck
Agents don't get tired. Your attention does.
Anthropic measured how its own staff work with AI agents. The agents now do far more hours. The limit is the person who has to read, check and decide.
An agent drafts the report, the code or the summary in minutes.
- Reading
- Checking
- Deciding
- Fixing
- Explaining it to others
Illustration of the idea, not a measurement.
“An army of Claudes, with every one asking for my attention.”An Anthropic engineer, quoted in the study
Source: Anthropic Economics team, How employment is changing at Anthropic, 16 October 2026. Employee-directed interactive sessions; human time counted as 40 hours a week per head. One company studying its own staff, who chose to adopt the tools.
Powered by Thorsten Meyer AI
Why Agent Updates Consume Human Time
The advice addresses a practical limit in workplace automation: more machine activity does not automatically mean less human work. When agents create frequent summaries, alerts and follow-up ideas, employees may spend time sorting the output instead of completing the work they delegated. That makes the design of reporting and review part of an agent’s real cost.
Meyer’s approach also offers a way to preserve human oversight without requiring people to inspect every action. Asking an agent to show its calculations or distinguish assumptions from known information can focus review on the parts where errors matter most. The recommendations are operational guidance, not evidence that a particular prompt or agent system will save time in every workplace.
Popular right now
Shark Rocket Pet Plus Corded Stick Vacuum, Bordeaux, HV322
Breville Bambino Espresso Machine, Brushed Stainless Steel
Breville Barista Pro Espresso Machine with Grinder, Brushed Stainless Steel
As an affiliate, we earn on qualifying purchases.
Anthropic’s Workplace Study
Meyer cites Anthropic’s research on its employees, published October 16. According to the figures in his article, by August 2026 employee-directed agents logged six times as many hours as the employees directing them. The study also found that agents spent about 80 percent of session time waiting for a person. Meyer reports that Anthropic estimated non-technical employees could spend about half their work time reading all the output generated by their agents.
Other figures Meyer attributes to the study point to increased coordination work. The share of sessions devoted to “meta-work,” described as mostly managing agents, rose from 10 percent in fall 2025 to 22 percent in spring 2026. Written standing instructions for agents increased from a few hundred to 8,500 over a year, according to the article. These measures concern Anthropic’s own workforce and should not be treated as a direct estimate for all companies.
Meyer notes that Anthropic’s researchers cautioned that its employees adopted the tools by choice and that the company’s working practices may not generalize across the economy. The findings offer a case study of attention costs, rather than a universal forecast of how agents will affect productivity.
“I made $20,000 in the last month and a half using Claude to do these professional proposals.”
— Dan Ninerell, owner of Modern Classical Chefs, speaking to Anthropic
How Widely the Findings Apply
The figures cited here come from Anthropic’s own employees, and Meyer reports that the study itself cautioned against treating the results as representative of every workplace. The article does not provide enough methodological detail to establish how agent use or waiting time was measured across different types of work. It is also unclear whether the recommended attention budget would suit teams with urgent monitoring or compliance duties.
The customer examples in Meyer’s article are attributed accounts, not independent evaluations. Ninerell’s reported revenue does not establish that Claude alone produced the result, and the source does not quantify the time saved or the costs of reviewing agent output. The article also presents a set of practical rules rather than a controlled test showing that they reduce workload.
Testing the Rules in Daily Work
Meyer’s proposed next step is for users to apply the rules to the agents they already run: identify the people responsible for reviewing and acting on outputs, set limits on recurring tasks, and require recommendations to include evidence and the consequence of inaction. Users can then check whether fewer, more focused messages lead to less review work without hiding important problems.
The source material does not announce a formal follow-up study or a new product release. Whether the approach improves productivity beyond Meyer’s own publishing and software work remains to be tested by teams using it. Any broader assessment will depend on how organizations measure the time spent producing, checking and acting on agent outputs—not just the hours agents run.
Key Questions
What is Meyer recommending people ask AI agents for?
He recommends asking agents to surface decisions that need human attention, with a recommendation, evidence and the consequence of doing nothing, rather than sending routine progress updates.
What did Anthropic’s study report about agent waiting time?
As cited by Meyer, the study found that agents in employee-directed sessions spent about 80 percent of session time waiting for human input. The finding concerns Anthropic’s workforce and may not apply to other organizations.
Why does Meyer say agent output can add work?
People still need to read, verify and act on outputs. Meyer argues that these human tasks can outweigh the time saved when agents produce frequent reports or recommendations that nobody has prioritized.
What are the four rules in the article?
Count the human work that follows; run small experiments without committing to every result; define stopping conditions before recurring tasks begin; and give a coordinating agent a limited attention budget.
Do the Anthropic figures prove agents reduce productivity elsewhere?
No. The figures describe one company’s workforce, and the article says Anthropic cautioned that its employees and work practices may not represent the broader economy. The findings highlight a possible attention cost but do not settle the effect on productivity across workplaces.
Source: ThorstenMeyerAI.com
