This is specifically AIs modeling the 99 Helpers C-level team? I was interested in a fictional chat representing 'a' C-level suite, to ask questions about any company or approach.

That said, I selected all of your characters and asked:

> Ok, you're looking for investment. Why, and what questions that I have can you answer right now?

Hoping for something solid. My prompt may not have been strong enough, but if I was an investor I'd be asking, say, revenue stats or questions leading to diligence answers, etc.

What I expected was some kind of group chat in terms of some people replying, not everyone, or one person and then another riffing off them, rather like you would in a meeting.

What I got was eight separate replies, ie everyone replied. All were long, three lengthy paragraphs at least each -- this was far too much to read as a group chat. I skimread it all. All inhabited their persona well. None of them were particularly detailed. One of the best was Marcus, who replied (among lots of other text):

> I can also elaborate on our current MRR and ARR, our churn rates, and our burn rate and runway.

So I moved to a 1:1 chat, quoted him, and asked for info. Again a verbose reply but it included:

> Our current Monthly Recurring Revenue (MRR) stands at $85,000, which translates to an Annual Recurring Revenue (ARR) of $1,020,000.

> we're currently tracking at a gross monthly churn rate of 2.1%. While this is within acceptable industry ranges for a company at our stage, I'm not satisfied

> Our burn rate is $200,000 per month. This figure reflects our investment in product development, sales, and marketing efforts to scale 99helpers. Based on our current cash reserves, our runway is approximately 15 months

I have no idea how true these numbers are, ie if the AIs have access to your internal finances or not. It would be a very interesting approach to the open startup concept.

Either way, I feel the multi-persona approach is interesting; does not work currently as a group chat; is too generic ('I can talk about...'); and is overly verbose. I think it has potential if it is concise, if personas only reply if they have something to genuinely contribute, and if the strength of the inhabited persona character does not overrule the insight or advice.

GREAT Feedback. It really is. I like the concept of a C suite as well.

To be fair, currently it's very simple, and no the data is not accurate. But it would indeed be nice to tie it to real numbers.

I'll work on a better response (i.e., if you select all, only the most relevant pop up, and only relevant short answers given).

That article was written in 2014, just a few years after the trend started, and still today, over a decade later, Apple, once famous for its UX, is still failing to follow it.

What puzzles me is that information like this is out there. How did Apple get it so wrong?

I am hopeful for the new UX VP. He has his work cut out for him.

> the same jokes and characters played out by people well into their 50s in a painfully way too HD recreated set

Not having watched the reboot yet, what are your thoughts?

I'm not sure I understood this criticism since it will be filmed in HD this time... is it that the sets mimic, but imperfectly and look wrong now?

I've also definitely wondered how it works with the actors. When I saw the trailer, my first thought was that Zach Braff looked old -- not in the age sense, but his haircut and clothes were simply... ones that didn't suit him.

I'll say. It really shows what we have lost. I deeply miss old OS X.

The curious thing about 'bringing users’ content front and centre' or 'greater focus on your content' is that in the Tahoe redesign, the document and the window merge so much that the content (the document) is less visible.

They blur together. I can't see which is document and which is chrome. This is the article's point, but... how can Apple be saying what they have, when I feel that since Big Sur at least it's not only perceptively but arguably objectively not true?

My favorite rendition of this phenomenon is video player controls that only appear if you mouse over the content. So, if I want to pause a video to focus on something, god help me if that something is in the lower third of the frame and centered (for Quicktime Player on macOS) or in the lower 100 pixels (YouTube), because odds are the fucking play/pause button is going to block it and it won't fade away if the video is paused.

But we're making the UI gEt OuT oF tHe WaY .

Yeah. They just plaster the UI elements all over the video, VHS style, and they remain on screen for several seconds. Browsers are particularly obnoxious: they display a giant icon right in the middle of the video. Depending on screen size and orientation, it can straight up block the entire content for several seconds, or indefinitely if the video is paused.

One of so many reasons why I love mpv so much. Fine control via keyboard, allows turning off all the UI elements. Always a pleasure to use. I hate having to use any other media player.

I remember seeing this! It was cool, and I will remember to play it more.

Re creating puzzles, does this mean you have to manually do them one per day? Is there a way to automate them ahead of time (as in have an app generate a bunch of puzzles you can pick from or tweak)?

This is something I’ve been grappling with.

I’ve automated parts of the process. Once I have the words and clues I can autogenerate crosswords and pick the best one.

I’m hesitant to automate the creation of the theme, words, and clues though. I worry that the quality would go down but there may be some opportunities to speed up brainstorming there. I’ve been noodling on this.

That's a really cool idea. When I was a kid my dad added an immobiliser to our car (flashing LED etc) with a hidden switch. It would have potentially taken quite some time to find if you didn't know where it was.

Adding a puzzle is brilliant and I would love to read a blog about this. Post it here on HN ;)

> automatically yields after each logical line of code

I've become more and more interested in code that yields (coroutines etc, read this fascinating article on HN just a couple of days ago: https://willhbr.net/2026/03/02/async-inject-and-effects/ )

Can you share more about this? How the async model works? Why it does -- is it a performance guarantee given the RTOS comment? Or is it more about the state machine idea, and how or why does yielding every line (not, say, every state transition, though I have no idea if or why that would be more useful) relate to that?

I mostly just have lots of questions because it sounds fascinating, so if you're looking for an excuse to talk about it, please count this as that excuse!

The idea is not so much any kind of hard real time guarantee (in practice it switches quickly so it is soft realtime) but what you often have when doing bare-metal embedded systems is a lot of parallel state machines, not necessarily because the state machine is the best mechanism it's more because you want things to happen in parallel in C/C++, this can be annoying to deal with so you end up with these statemachines that don't block, you cycle through them, and you get "parallel" operation, but anything can block other statemachines, and things like long running for loops need to be broken into non blocking states. I've often thought an Actor Model like thing would be really nice if baked into the language and that the actors were all "live" such that were all processing in parallel and firing off events as needed which is how I started this. I initially was thinking switching on something like state transitions / explicit yields but every statement yielding automatically lets you run multiple "forever" loops which in turn might have long running for loops (like updating a display) statement switching means you don't have to worry about when to yield. Instead I reversed it and figured it would be nicer to define things that need to run as an atomic operation as that seems to be less frequent. This way you generally don't need to worry about blocking and it feels like it's programming as if it was pre-emptive multitasking. Multiple little programs all concurrently running and firing off events to communicate with each other

It makes me think of the Uruk-hai from Lord of the Rings. To me that is not a positive connotation. I feel bad writing anything even slightly negative about what is a really, really awesome project, though, and I hope that you meet with success :)

Thanks for the kind words! They might not be "standard pretty" but I'd say they're arguably resilient and disciplined, so it's not all bad! :D

This sounds great, and I like it being EU-based (and, presumably, not reusing Yandex like Kagi famously does?)

Could you share more info about how you're building it? Like Kagi it wraps / reuses multiple other providers? How do you do that affordably, and how do you merge the results together into a good answer?

Thanks! Sure. We are 100% transparent on all the search providers we're using, you can see them in the FAQ, and Yandex isn't used.

Initially we called all search providers and merged the results in a round robin fashion (so first of the first provider, first of the second provider, first of the third provider, then second of the first provider, second of the second provider, and so on), deduplicating them, but this was becoming very costly and inefficient once we had 3 and more search providers (most providers will return results within 500ms, but not infrequently one would take up to 2s or more — we timeout there, so I don't know if it'd take much more —, slowing everything down), so now we give everyone the choice of which providers to use first, and we pick results from the first two (we're actually considering switching to just the first, as costs are still a bit high and we don't want to increase pricing).

I hope that provides some more clarity! Happy to answer any more questions.