Build core skills to thrive as an AI-era developer

still don’t know how they’re doing it

The two people transition was not needed for this talk, it just increased the cognitive context switch load to consume this

im learning

Come back to this talk in a year and see how trivial the whole talk will sound. Don’t delegate judgement they say now, that will change quickly.

Anyone else noticed since this agentic approach started being pushed the amount of critical and high vuln CVEs and service interruptions have gone on the rise across most big companies

The term agent is thrown around like a marketing buzzword, and to a large extent it is. Agent has a specific use case and outside of that use case you should avoid it like the plague. And that use case is whenever you are dealing with unstructured data.. eg natural language in an email or a web page or something typed into prompt, etc. An agent is not a general purpose automation tool like it’s being sold by the AI companies. If used for anything other than processing natural language (via llm) agents will create new problems like cost, uncertainty, slowdown, etc.

Would be nice to see more concrete examples, not just bullet points

Bla bla bla. No real advice here.

these presentations feel like water

I want to work with Google someday. Not just for the name, but because I want to become the kind of engineer capable of operating at that level.

Untill intend is also automated😅

lost me at "top performing engineers" and "engineers that use more AI spend more time ideating". what is a "top performing engineer"? what's your metric for that? and the second point is really just a positive rephrasing of the fact that coding agents are dopamine-driven slot machines similar to social media. of course they tend to increase engagement, that is not necessarily a positive thing at all, as some research is already starting to indicate this can be very bad for mental health. also, there is research (harvard business review specifically has a few critical articles+data) contrary to many of the points they make, for example workers using gen AI to handle work that is adjacent to their core competence tend to cause more downstream work for people with that core competence (ex.: producing garbage code in a codebase they know nothing about, ...). i've seen it first hand, and i'm sure many of you have, too

I thought spec-driven development is a 2026 Q1 thing? We are in Q2, nearing Q3. The problem with spec-driven development is that human defined spec isn't enough for the AI to carry out its job. For instance, the constraint of keeping API response remain <200ms under heavy load, what if it is impossible and some compromises had to be made? Will the agent goes against the Anti-goal of not sacrificing relevance for speed? And what about side effects? You can't explicitly write that because you might not actually know it (else it won't be a side effect). I think Google is able to pull this off not because of a specific method or agents but the perfectly engineered system that was built over the decade. The question is, how long can they sustain until everything collapse?

Out of touch and boring.

T-shaped staff have been desirable for ~20 years. Agile encourages generalists over specialists and this talk sounds like agentic AI reinforces this.

😍

This was a very packed Exec talk to introduce consulting or that the business is going the right way. Would love talk series on the getting to these points or practical steps. Almost every slide needs a talk to explain it. (It’s great google can but how can other companies unpack it too)

Here’s another common truth about ai - use the foundation of delegation and treat ai as a junior or a trainee.

So nothing changed, T-shaped engineers are the alpha as for the last 10-15 years.
I wonder, who were those thousands of devs google has laid off. Weren’t they t-shaped enough?

Great video! 😊

Why they measure progress on number of code lines?

I really love this video, anyone can share the slides please

Great content. Thanks for sharing this

Wow, just loved the video, great explanation., thank you for sharing this Google :) 👍

I'm sorry but director level people are always so removed from actual work, even if their job is to study how the work is done (like the second speaker). It is inevitable. Much better to ask the person on the ground doing the work or their direct supervisor.

The last two points, changing team typology and operating principle (what we care, provide psychological safety) are the hardest part. Human nature, especially within incumbents there is always inertia to change

really good, but too basic, i want to see real solutions.

I stopped watching after the T shaped talk

Don't you think guys, this clarity is crucial?

VALUABLE

"I know Kungfu".

So we are rediscovering waterfall?

Thanks for this master class about software engineer with IA. The concept about the T was very important for me

I thought this is about SKILL. 🤣

come on the T shaped stuff is so tired

This was exactly what I needed, thanks!

10th... Thanks