aenis

Yeah, I know -- though I think for the time being the screens-as-mirrors extend to cars that would otherwise lack a sensible center rear-view mirror, so thats a net positive.

On screens replacing side windows, I don't have first-hand experience. I see them on trucks and some fancy Audis -- but since I know for a fact that Audi designers are dangerous, well organised people trying to kill me with their safety features, I avoid their cars religiously.

My argument was to keep the attention of the driver on the road. To that end, I don't care whether I am looking at a physical mirror or a screen that replaced one (as long as its good, high res, low latency screen) - I am looking in the direction I am driving and focused on steering the vehicle.

For the same reason I don't mind (in fact, I appreciate) the silent helpers such as ABS, ESP, 4x4 and so on - all of those systems exist, work, and never utter as much as a beep to distract me. Great.

Popups, imbecillic charging/energy distribution animations, elaborate sequences needed for basic functions such as AC controls, are the things I don't like. Sure, people designing this stuff should engage in research, but some things are actually obvious. Such as the need to mind the mirrors when reversing.

True, though on the volvo its the defaut - when you hit the cruise control button, it automatically enables both the lane keeping and the adaptive cruise control. Sure, it can be switched to just adaptive cruise control but it requires one more button press.

Likely.

What does this sentence even mean? "if you want to connect, you have to make the magic work behind the screen". It crashes my parser. Good thing I am not reading hacker news while driving :-)

Yeah, my volvo also wants me to do questionnaires when driving. Insane.

When I am buying a new car, I now always try to rent one, and specifically the current model year, for a few days and do various types of driving. My V60 used to spend some time in the garage and I got various new models as replacements. The new one, for instance, has a choice of two behaviours when it thinks you are above the speed limit:

- beeping - or, in order to speed above what it thinks is the limit one needs to release the throttle and press it again

The main problem of course is that its very often mistaken about the speed limit.

Another problem. The thing recently got a new major version of the infotainment system. On my 2 year old V60 it is now noticeably more laggy, for instance when bringing up the AC panel its at least 1.5 seconds before it comes up. Now what is more likely - that I will press the button and regain focus on the road, or that I will press the button, and be distracted for a second or two longer?

I used to drive a range rover sport that would display a long pop up with some legalese about focusing on the road while driving when I hit the navi button. It required acknowledging.

"So, that's attractive as a slogan but it's 100% incorrect in practice. Non-road UI features like backup cameras and blind spot warning alarms save lives. Period."

The "on the road" extends to mirrors (or screens that have replaced it) - I assumed that was obvious.

I had the same thing while passing an unexpected cyclist. That was an Audi A6, I vividly remember even though some 5 years have passed - I was one of the scariest things that happened to me on the road.

Call me old school, but I'd like a physical master switch to disable all of the systems you mentioned. I drive a lot, and often in rented cars, and in various countries.

- automatic braking - i brake gently and then do a limousine stop. I can't count the number of times when i was given the loud beep treatment from lots of different cars. I never rear ended anyone in about 1.2m kms driven.

- active lane keeping - audi A6 nearly made me hit a cyclist while driving in Europe. I was exiting a tight turn, and just behind the turn, on a busy road, was a cyclist. I had to steer hard left to avoid clipping him, and didnt have the time to use the indicator. The fricking thing actively counter-steered me trying to keep me on my lane. Incidentally no automatic braking at the same time. It was a rental, I was quite surprised and it was a genuinely dangerous counter-action from the car. No thanks.

- smart cruise control. Nice when it works. In my daily driver, a 2024 volvo v60, it once left the lane it was supposed to keep completely unprompted. Good thing I was holding the steering wheel firmly. No thanks.

- lane change alerts - nice when done right. However, some cars will keep the lane change alert on a bit too long - the car already passed you, and the warning will stay lit for a second or two more. Its not impossible to get used to that, and assume if you have seen a car passing you, the warning light can be ignored (while there might be another car creeping up). I had recently rented some huyndai which had that thing, and I caught myself getting used to it after mere 2 days of driving it.

- rest breaks - i think i had this on a rental huyndai. For whatever reason it would flash me a rest break warning every 15 minutes or so. No clue why, I wasnt driving for more than 1hr, and was completely rested. It was distracting me with that stuff for most of the journey. No thanks.

I genuinely like ABS, ESP and thats about it. Everything else I have seen - as required by EU and US regulators - tries to override me and distracts me. As I am getting older, I am less and less tolerant of distractions.

Yes, you are right. All my power tools (from Makita, Milwuakee, Festool) have absolutely phenomenal UI - so there are still corners of industrial design where the dark patterns/attention grabbing product people haven't ventured. They should be brought back to car companies.

I really don't know about that. Mercedes used to mean high, tactile, audible mechanical quality. You'd hear it while closing the doors, you'd see it when looking at perfectly assembled dash, you'd hear it while driving and you'd be happy with it when clocking 500k miles with just regular maintenance. I remember those cars - I think the quality started tanking around late 1990s.

Right now its just ok. My friends S class has visibly mis-aligned buttons (a 200k car). My other friends electric S-class bean-thingy has squeaking doors (a 2 year old, 120k-when-new car) and feels surprisingly cheap to touch and drive. Sure, small sample and all of that but I don't think those are exceptions.

I only drove one Chinese car, and it was just a normal experience - what I'd expect from a volvo, bmw, or audi. Good UI on the infotainment, was below average annoying. No big difference vs. a merc. For sure not a qualitative difference in levels of refinement.

'He also explained that "I'm a big believer in screens, because I really believe if you want to connect, you have to make the magic work behind the screen." '

I am a big believer in keeping "product people" away from UI design for dangerous machinery.

The eyes and the attention of the driver should be on the road. All the audio visual noise from the car is just plain dangerous. I don't want my car to draw my attention to itself for anything less than a critical engine/tyre pressure failures. I do not want beeps on anything else distracting me while I am driving.

My Volvo will, for instance, flash the same type of visual alert when fuel level is low (permanent "do you want to navigate to a fuel station" modal window obscuring navigation, speedometer and so on) -- as when it encounters a serious engine malfunction. It will steal a bit of my attention when it pops up. One of those days, someone will have an accident because of this moronic design, its statistically certain.

Same with wipers fluid level low. I need to click on the button to hide the message.

It will on occasion beep very loud when it thinks I am not braking hard enough. The map in the google android car navi rotates when i am just trying to pan. When I want to select an alternative route I need to very precisely touch a very small area on the screen, and more often than not instead of selecting the alternative route it will actually rotate the map.

It is clear to me that either the people designing car UIs are staying away from those cars, or are just incompetent. (Or, I guess, both).

Likely, and via vertex on gcp (or whatever they are calling it this year).

Which also means, if you are a big boring AWS or GCP shop, and have a spend commitment with either as part of a long term partnership, it will count towards that. And, you won't likely have to commit to a spend with OpenAI if you want the EU data residency for instance. And likely a bit more transparency with infra provisioning and reserved capacity vs. OpenAI. All substantial improvements over the current ways to use OpenAI in real production.

You'd have learned something new. Useful, not useful, thorough understanding of a new thing is rewarding.

Also, its not primarily about money - the real investment here is time.

For sure, but these days product management mistakes can be more easily rectified. Before that, if we invested 4 months in building something that did not land, we'd be quite reluctant to jettison this and start fresh. Egos, career considerations, sunk cost, etc. I think I will soon be able to say "not any more", since doing a U turn can be cheaper than pretending the bad choice is the best choice. "Oops, lets redo this" vs. 6 months of executive squabbling about whose fault it is that we wasted $3M in development costs on something that clearly does not perform.

Also, give it time. Real adoption in boring companies started Q1. Q2 is, I think, this settling in and people learning how to do their work and manage their responsibilities. Q3/Q4 will be the time when I expect to start seeing higher velocities across all IT-adjacent products I use.

As a CTO I can say that this is not my experience.

My experience these days is fighting corporate bureaucracy and inertia to make sure we reap the benefits of faster coding. Feeding agents with work is not a problem. Building teams that use those tools effectively is the problem. (Say, shall we merge product and engineering teams? Do we start getting rid of people who refuse to use AI? What do we do with pentests? How do we strengthen the tools that do code analysis and weed out lazy devs who can now more easily pretend to be invested in their work? Stuff like this keeps me busy.

The people who use AI to the maximum learn more.

A neutral hobbyist on a $20 budget will build something and immediately bump into quotas. Its not going to be an enjoyable experience.

A negatively predisposed pro who only dabbles in AI gets to the first disappointment, smiles, and thinks "yeah, about what i expected" and quits.

To learn those new tools one needs to not be stingy. Invest as much as needed into tokens, subscriptions, and maybe most importantly invest the time. Spend time building various things. Try out various models not just for coding, but as part of apps being built. For bonus points, meaningfully experiment with local models. I try to avoid discussions with sceptics who have not put at least a few months of effort into learning those tools. It's like discussing driving with my mother in law, who spent maybe 20hrs behind the wheel through her whole life (and is very, very opinionated!).

In a word, bottlenecks moved.

What I see in my backyard: coding now takes significantly less time, but its just coding. Before one gets to building there are squabbles between business and product people. Testing takes just as much as it used to. Since nice to haves are easy to add and product people begin to take it for granted, the product cycles don't get shorter.

Give it time. Right now its just coding, but procedural AI will come after product development, architecture, and then whatever is left of management.

"You are absolutely right!"

My experience as well, unfortunately. I am really looking forward to reading, in a few years, a proper history of the wild west years of AI scaling. What is happening in those companies at the moment must be truly fascinating. How is it possible, for instance, that I never, ever, had an instance of not being able to use Claude despite the runaway success it had, and - i'd guess - expotential increase in infra needs. When I run production workloads on vertex or bedrock i am routinely confronted with quotas, here - it always works.

Thanks! In our limited experience, Claude does not focus that much on guardrails and code quality when building a feature - but can be pretty focused on code quality and architecture when asked to do just that. So, one a few hours to iterate a feature, a few hours to refactor. Rinse and repeat.

I am not sure how others are doing this, but here is our process:

- meaningful test coverage

- internal software architecture was explicitly baked into the prompts, and we try to not go wild with vibing, but, rather, spec it well, and keep Claude on a short leash

- each feature built was followed by a round of refactoring (with Claude, but with an oversight of an opinionated human). we spend 50% building, 50% refactoring, at least. Sometimes it feels like 30/70%. Code quality matters to us, as those codebases are large and not doing this leads to very noticeable drop in Claude's perceived 'intelligence'.

- performance tests as per usual - designed by our infra engineers, not vibed

- static code analysis, and a hierarchical system of guardrails (small claude.md + lots of files referenced there for various purposes). Not quite fond of how that works, Claude has been always very keen to ignore instructions and go his own way (see: "short leash, refactor often").

- pentests with regular human beings

The one project I mentioned - 2 months for a complete rewrite - was about a week of working on the code and almost 2 months spent on reviews, tests, and of course some of that time was wasted as we were doing this for the first time for such a large codebase. The rewritten app is doing fine in production for a while now.

I can only compare the outputs to the quality of the outputs of our regular engineering teams. It compares fine vs. good dev teams, IMHO.

That.

We already shipped 3 things this year built using Claude. The biggest one was porting two native apps into one react native app - which was originally estimated to be a 6-7 month project for a 9 FTE team, and ended up being a 2 months project with 2 people. To me, the economic value of a claude subscription used right is in the range of 10-40k eur, depending on the type of work and the developer driving it. If Anthropic jacked the prices 100x today, I'd still buy the licenses for my guys.

Edit: ok, if they charged 20k per month per seat I'd also start benchmarking the alternatives and local models, but for my business case, running a 700M budget, Claude brings disproportionate benefis, not just in time saved in developer costs, but also faster shipping times, reduced friction between various product and business teams, and so on. For the first time we generally say 'yes' to whichever frivolities our product teams come up with, and thats a nice feeling.

Boring corporate Ai will surely come, but hey, lets enjoy the wild west while it lasts. I am grateful to see Boris come here to address problems people face. I 100% sure nobody is making him - he has one of the coolest jobs in the world.

This, 100x.

I do M&As at my company - as a cto. I have seen lots of successful companies' codebases, and literally none of them elegant. Including very profitable companies with good, loved products.

The only good code I know is in the open source domain and in the demoscene. The commercial code is mostly crap - and still makes money.

True, but its really hard to name a family of commercial devices with security features in hardware, including serious security features, which were not eventually hacked.

Worse still, for new mainstream devices that are believed to be safe the state sponsored actors will likely operate unpublished exploits, and will exploit the misplaced faith people and judiciary will put in device attestation. I dont think the very likeable people who worked on Pegasus found themselves respectable jobs - they are likely still selling that sophisticated crap to all authoritarian regimes.

Exactly this. And whats more, the idea of device attestation makes people trust those devices, and the history of rooting consoles and phones proves that nothing holds, even tech backed by billions in commercial interest.

The whole point in reducing the blast radius is valid - by all means make this optional and allow the user to elect to tie their identity to the device. For everyone else, implement validation of actual transactions, not just user secrets and device secrets.

Fingers crossed for the judiciary - if the implementers ignore the intention of the law, then lawyers will have to help them understand the limits of corner cutting - and block this.

Step by step. We realize we will not get there in one day.

Its the same as with bicycle paths. Initially - those make no sense, leading from nowhere to nowhere. Give it a few years, and a usable network emerges.

Right now there is serious money and brainpower being poured into sovereign cloud tech. Thanks to the gift of open source and standards, its actually not impossible to create modern systems with zero US dependency.

I fear, though, that as with everything else Microsoft Excel will be the hardest dependency to deal with.

If this was a gym subscription, it would be an equivalent of some people going to the gym, and some people sending their android to the gym every day, for the whole day, and using as much equipment as the gym policy allows.