Tech Talks by Thoughtworks

***Register here: https://www.linkedin.com/events/7482324853468090368/***

What if every argument about agentic engineering — dark factories, code review, model steering, production trust — is actually the same argument in disguise?

It is. And the question at the centre of it is: how much autonomy do we hand to an agent, and how do we stay confident in what it does?

Join us for a LinkedIn Live session on 22 July with Kief Morris, Distinguished Engineer at Thoughtworks, as he shares a framework to reason about that deliberately — not argue about it in the abstract.

We'll cover:

• Why line-by-line human review never was the guardrail we told ourselves it was — and why it can't survive the volume agents now produce
• How to define the "unit of work" you hand to an agent: its batch size, its span across the delivery lifecycle, how you prepare and verify it
• Where the rigour we used to spend reading code actually goes — upstream into intent, and downstream into an automated harness
• A practical decision model for setting agent autonomy in your context, based on verification cost, cost of being wrong, and how well you understand the requirements

If your team is shipping with agents and finding it hard to agree on how far to trust them — this one's for you.