Discussion

Monks says its AI agents now work for days without supervision

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#4581

S4 Capital’s Monks says its AI agents can now work on coding projects for 36 to 48 hours without supervision, checking their work as they go. The agency is also adapting the setup for marketing, putting a live example of how agent autonomy is changing workflows, budgets and human roles.

Watch Desk analysis

What happened

Monks chief AI and revenue officer Wesley ter Haar told Digiday that agents divide coding projects, test their work against requirements and review one another. He said agents can now build their own multi-agent setup, where staff previously had to write detailed instructions for how the agents should co-ordinate.

Monks is adapting the approach for social and performance marketing, where agents could respond to trends, comments or posts that are underperforming. Ter Haar said an unnamed client is expected to move social work onto the system early next year, from insights through to community management. He said the client’s contract and oversight requirements will shape what the agents can do.

Why it matters

This is a specific account of agents moving beyond short coding tasks into longer-running work and marketing workflows. It also shows the less glamorous questions that come with the extra autonomy: who sets the brief, who checks the output, and how much human time should still go into improving it.

Ter Haar said agents can generate dozens of ideas quickly, but people still need to judge which are worth keeping. He also described a shift in budgeting, with the company weighing token costs alongside staff costs. S4 Capital has cut about 800 jobs over the past year, according to the article; the account does not establish that agents caused those cuts.

Our read

The striking detail is not simply that agents can run for days. It is that Monks is trying to turn that endurance into a working process, while still relying on people to set requirements and choose which ideas survive. Autonomy may save hands-on time, but it does not make accountability or judgement disappear. It may just move them to a different part of the day.

What to watch

  • Whether the planned client deployment goes ahead early next year, and what oversight it requires.
  • Whether Monks shares measurable results on quality, cost or time saved in marketing workflows.
  • How the company balances staff time against model-token spending.
  • Whether longer-running agents change how teams organise work, rather than simply speeding up the old routines.

Discussion spark: If agents can run for days but people still have to set the brief and choose the winners, should businesses count their cost against software budgets or staff budgets?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.