alex kerss

Phd in mathematics from Cardiff University UK, property developer turned app developer

[alex kerss @jungleskellam]

@tonysimons_ Let me know when another TestFlight is available and I’ll give run through it again.

[alex kerss @jungleskellam]

@thegeorgerosu @rileybrown I’m in the UK, I just downloaded it by using a VPN and I was using it. Haven’t tried turning off the VPN yet but I’ve got it. I’m in the UK

[alex kerss @jungleskellam]

@milindlabs the the kindle app is coming when?

[alex kerss @jungleskellam]

@Muse when will there be a kindle version, i need to control my muse bot on my kindle

[alex kerss @jungleskellam]

What’s the last thing you built with AI that anyone still uses?

[alex kerss @jungleskellam]

@RomainDerrien So essentially, you’re saying, it’s not possible to achieve AGI in terms of AI writing articles as well or better than a human can this will never happen in the future

[alex kerss @jungleskellam]

So what happens, when AI can write articles that are no longer slop and have an higher engagement level than a human article?

Like at that point do all human posts on social media become superseded and no longer show up?

And surely, we are only like a couple of years away from that, or is that just completely impossible, and that will take decades or centuries, or that could never happen?

[alex kerss @jungleskellam]

Anyone else want an Opus 5.5 review that isn’t a game?

[alex kerss @jungleskellam]

Codex Usage Reset FOR ALL in 43 minutes, just hang on, they need to wait for my weekly to reset first...

[alex kerss @jungleskellam]

@kelvinbuildss BBC BASIC

[alex kerss @jungleskellam]

Codex Usage Reset coming in 3h and 27m, they are just waiting for mine to tick over first

[alex kerss @jungleskellam]

@joyrap What does that mean?

[alex kerss @jungleskellam]

OpenAI will be practically gone in 10 years.

Anthropic will be practically done in 10 years.

I don’t see how they can survive.

Here’s why…

AI intelligence from open source models in 10 years’ time will be more than sufficient for 99% of use cases. We’re talking text, image, video generation. So it comes down to selling your platform.

No one’s gonna pay $200 a month for the Claude or ChatGPT desktop app when models that are ten times more powerful than today’s models, that run at lightning-fast speeds, will come preinstalled on even the cheapest of laptops or phones. These models will be free, open source models.

The cost for OpenAI and Anthropic to develop such models just burns more cash than they can ever get back selling subscriptions to their platform. Because remember, the platform is no longer the model. It’s just the platform. It’s the harness.

The platforms that will be able to sell are ones like Muse, because their platform will have access to the whole ecosystem of Facebook and Instagram. Facebook and Instagram will restrict other harnesses and make you want to pay $20 a month for Muse. Grok bot and X have a similar play, but it’s not as strong.

The real winner will be Google. Their personal assistant that looks like Grok, that looks like Muse, will have access to the whole Google ecosystem, which is probably more valuable.

Facebook and Google are the winners. No one’s gonna buy a ChatGPT subscription to a platform that offers you nothing compared to anyone else.

Building a platform with all the features is not the moat. Coding agents will be so capable that building the platform with the full features is not the moat like it is at the moment.

People will pay for Codex now, even if you’re not a coder, because the Codex application itself is so good. That won’t be the case in 10 years’ time. There’ll be free applications like that, that use free local open source models.

You won’t pay just for the application. You’ll pay for the ecosystem it taps into, which is why Google and Facebook win.

[alex kerss @jungleskellam]

Thats interesting reading your take and i agree with you that they are moving to these new work surfaces, im just not convinced at all that revenue generated form those with the competition that will compete against for those surfaces will fund their massive compute costs to develop frontier models, and its those frontier models that essentially makes them held in the regard they currently are.

[alex kerss @jungleskellam]

I've been developing my product since March. My platform is essentially the same as Grokbot and Muse. When Buzz came out, I was a little sad. I was completely naive, thinking, “Why has no one else thought of this?” When Grokbot came out, the number of posts on X from other people who had been developing an agent platform like this made it seem obvious that everyone had this idea.

Can we compete with Grokbot and Muse by releasing our own versions? I don't think so, not in terms of the raw numbers. Do we really need to? If, for example, I built my own agent platform and got 2,000 users paying a subscription, that would probably mean 0.000001% of the market is available. That's probably all I need as a solo developer.

I think it's more about whether you can find an audience. I certainly have no audience following me, no audience trusts me, and no audience would consider subscribing to any product I released. If I built up a good audience, I think I could develop a Muse or a Grokbot and sell it as a SaaS product.

I think it's not really about going head-to-head with these big tech companies, but about building up an audience big enough to make your product viable. That's the way I'm looking at things now.

[alex kerss @jungleskellam]

Microsoft got embedded when competing products were taking years to build. Anthropic and OpenAI are getting embedded into enterprise, but whether that embedding can survive the new era of rapid software engineering, I’m not so sure.

With Microsoft Office, the cost is not that much, really, per person. If OpenAI and Anthropic can only charge that same cost, assuming they can even stay embedded, it’s just not going to cover their costs for developing frontier models.

They’re just not that embedded yet either, so they’d better quickly get mass adoption, like Microsoft did, or they’re gone. Because enterprise, yes, is definitely the real market. I just don’t see them getting embedded like Microsoft did, not in the days where you can just build these products so quickly.

[alex kerss @jungleskellam]

@sattyyouneed no, ive been trying it, complete waste of time

[alex kerss @jungleskellam]

@cynthiamcgillis Muse is INSANE

[alex kerss @jungleskellam]

@jiang_kev @PolarBrowser so how is this better that codex with broswer and computer use?