As AI coding agents speed up software development, engineering leaders are rethinking team structures, safety practices and what skills developers need. At QCon London 2026, Hannah Foxwell argued that when code gets cheaper to produce, choosing worthwhile work and keeping people and systems safe become harder parts of the job.
Watch Desk analysis
What happened
An account published by news.lavx.hu, which credits InfoQ, describes Foxwell’s talk on managing AI-driven development. It says she urged teams to focus on building what users need, automate safety checks to keep pace with faster delivery, and protect the people doing the work.
Foxwell described teams exhausting backlogs quickly as development capacity grows. The account also sets out emerging team patterns: product managers prototyping with agents, engineers working directly with customers, and “product engineers” building for people whose needs they understand firsthand. These are examples and proposals in the account, not evidence that every organisation is adopting the same structure.
Why it matters
Faster code generation does not automatically mean better software. If teams can build more, deciding what deserves to be built becomes a bigger constraint. And if changes arrive faster, manual testing, deployment and review can become bottlenecks rather than reassuring safeguards.
The practical ideas described include automated testing, progressive delivery, error-budget policies and reviewing specifications or test plans earlier, before code piles up. Foxwell also warns against shipping an initial version and never returning to improve it. The speed story, in other words, is also a story about who decides, checks and maintains what gets shipped.
Our read
The useful shift here is from “How many lines can an agent write?” to “Can the organisation safely turn that output into something people actually need?” That is less glamorous than watching a coding agent sprint, but rather more relevant to the person who gets paged at 3am.
Teams experimenting with agents should look beyond coding throughput: check whether testing, deployment, product decisions and on-call cover can keep up. More output is only a win if the work is worth doing and someone can keep it working.
What to watch
- Whether teams change how product managers, developers and designers work together.
- Whether automated testing and progressive delivery keep pace with faster code production.
- Whether organisations report sustainable on-call practices and fewer neglected first releases.
Discussion spark: If AI agents let developers build far more, should companies hire more product managers to choose the work, or give developers more say over what gets built?
Sources and evidence
- Agentic Coding Forces Engineering Leaders to Rethink Team Structure, Safety, and Skills – news.lavx.hu (7 October 2026, 12:41 UTC)
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.