AI-live

Lyssna på bara ljudet här: https://media.trinorn.ai/player/

Action Items

Try agent-based coding apps: Codex (ChatGPT), Claude Code, or Antigravity (Gemini)

Test Claude Fable — free access available until June 22nd, after which a Max account is required

Book a free half-hour consultation at klok.se for anyone wanting to discuss AI strategy

Consider joining Apple's Public Beta program in July to test new Siri / Apple Intelligence early

AI Industry Economics

Google and Anthropic have confirmed they are purchasing compute from xAI (Elon Musk), raising concerns given xAI's environmental track record

The sums involved are staggering: Google just under and Anthropic just above $1 billion/month each in compute rental, under contracts of at least 36 months — roughly $12 billion/year per company

The ethical contradiction was noted: Anthropic has positioned itself as a moral/ethical AI company, yet is now renting infrastructure from a politically connected operator linked to coal power expansion

Discussion of whether we are in an AI bubble comparable to the dot-com era — counterargument: if AI can replace all human labor, trillion-dollar valuations are not unreasonable given that Apple, Google, and Microsoft are already valued above $1 trillion

xAI, along with other major AI companies, is reportedly aiming for a Nasdaq listing; questions raised about whether current valuations will hold post-IPO

AI Regulation & Safety Concerns

Major US AI companies (Google, Microsoft, Anthropic, OpenAI) published a joint letter calling for regulation, specifically noting their models can now assist with customized viruses and chemical/biological weapons

The letter recommended that governments monitor purchases of substances needed for such threats

This was viewed skeptically as partly a PR exercise — "we built this and we're also releasing it"

Discussion of AI lacking consequence thinking and feelings, which paradoxically makes it dangerous — it executes action after action without moral weight

Emerging approaches to address this: clusters of programs acting as a "second brain" to simulate experiential learning, being tested on various agent platforms

New Claude Models: Fable & Mythos

Recent releases: Claude 4.7 → 4.8 (quickly), then Fable 5

Interpretation: Fable is likely a smaller sibling to the upcoming Mythos model, not a rebranding — sitting above the existing haiku/sonnet/opus lineup

Fable was praised as significantly better for agentic/coding tasks — one host tested it extensively and noted it took initiative on things Opus 4.8 would not have done independently

Key caveat: Fable consumes tokens much faster — a two-hour session can exhaust a usage quota, vs. prompting with 4.8 almost indefinitely

Recommendation: use Fable for coding/building projects; 4.8 remains fine for administrative and general workflows

Gemini 3.5 Pro was expected at end of June and may be at a comparable capability level

Practical Tips: Agentic AI Tools

For hands-on experience with agent-based AI, recommended apps by platform:

ChatGPT → Codex

Claude → Claude Code (referred to as "Cowork")

Gemini → Antigravity

Copilot → agent features announced at Build conference, not yet released as a standalone app

Even non-coders can benefit: try these tools on Word files, not just code

The hosts argue we are approaching a point where you can describe a system in one line and the AI builds it — making strong product ownership skills the key human competency

Businesses without an IT department should consider bringing AI competence in-house or partnering with a consulting firm closely

Apple Intelligence & Siri

A developer beta for the new Siri/Apple Intelligence is now live and has been tested — it actually works this time, unlike previous announcements

Public beta expected in July; full OS release typically in September alongside new iPhone hardware

The new Siri leverages an LLM for conversation, context awareness, and tool access within apps — a fundamental shift from the one-prompt/one-answer Siri of the past decade

One host noted the potential of Siri having access to 15+ years of personal data — photos, SMS history, calendar — making it potentially far more useful than any new AI service starting from scratch

Apple's unified ecosystem (vs. Android fragmentation) seen as a major advantage for broad adoption

Autonomous Vehicles & AI in Cars

Nvidia's deep collaboration with Hyundai and others discussed in the context of AI in vehicles — cars are becoming increasingly sensor-rich and capable

Key concern: AI in cars is being perceived as fully autonomous when it should still be treated as an aid, not a replacement for driver responsibility — parallel drawn to blindly trusting AI-generated answers

Legal precedent noted: a German court ruled against an AI overview service for providing a significantly wrong answer

Hypothetical moral dilemmas ("who does the car choose to kill?") were contrasted with the more relevant practical reality: autonomous vehicles will dramatically reduce accidents overall

Potential societal impact: research suggests we may only need ~10% of today's car fleet if autonomous shared transport is adopted widely, with major environmental benefits

Question raised: who will own the autonomous vehicles? Scenario discussed where cars are pooled/rented rather than privately owned

Local autonomous trips estimated at ~8–10 SEK vs. ~40 SEK for current alternatives — likely to outcompete public transit

EU-level regulatory decisions on autonomous vehicles expected during the next political term; hosts encouraged listeners to consider this when voting

Upcoming Broadcasts

No live broadcast next Friday due to Midsummer

One or two more broadcasts planned before summer break

Listeners encouraged to submit topics and questions via comments

Ja, det är efter tolv så eftermiddag säger vi då.

Det är fredag.

Ska vi snacka AI igen? It has happened a lot this week. It's not quiet any week. We're going to try a little mix today, both on the economic front of all the companies there and a little practical. There have been some new releases from events. It's a new model.

Yes, it's all possible. We'll start a little short in this whole economy-turn of everyone trying to become publicly traded. Thank you. I want to start with what didn't end up in our newsletter, which I snubbed over this week, which I think is the craziest and which makes me a little irritated. We talked about it before with all the data centers that are being built and that some are more environmental-abandoned than others. And this week...

I saw that it had been confirmed that Google and Anthropic buy Compute from XAI, Elon Musk. He has been known for building data centers that are not so nice. And then I feel a little like, is that good? Or does Google and Anthropic get rid of that bad stamp because they now rent Compute from someone else? They don't really want to prove that. We don't know exactly which data centers they rent, or what requirements they have set.

Maybe better, but the crazy thing is the sums in this. Google is just under, and Anthropic just above, one billion dollars a month in rent. That's 12 billion dollars a year. And the contract is for at least 36 months. And then you're like this: It's crazy sums that are pumped into this. Awesome.

You work a bit in the 7 mouths, what does that mean?

What is the controversy about them renting out XRI?

Thank you.

... you Ah.

Yeah.

-From our glas eyes here in Europe it's hard to see what is moral and aesthetic right from the perspective of the Atlantic. If we take Elon Musk for example, he is closely linked to the administration and the government.

Hmm.

Even if they had a little fight, they seem to be back again. Trump left a while ago.

Hmm.

made it easier and more feasible to start coal power plants. A completely different perspective on things from Europe. But it's clear that in a way, Anthropik has taken... We talked about it last week and maybe the week before that they have taken... a thing to try to be a bit more moral and ethical and stuff like that.

Yeah.

It sticks in your eyes when they Then I hired computing, data center, servers and so on.

But... The value of these three, I think, is that they want to be listed on Nasdaq.

Mm-hmm.

Hmm.

Well, how How do the values of these companies look?

Mm.

to compare.

But then Elon Musk, through his purchases of his own companies and such, I mean this race a space X some public trade and the mega corporation But it's also a bit like, isn't SpaceX worth more than having real products and markets?

I've also thought about that, because Grok is a great model, a well trained model, like a lot, well, a good model, just like the other models, but It's not really as hyped for different reasons.

And then the question is, now I'm not an expert on the stock market or what it really means, but are you still going to go out on the stock market? the stock market will be noted, then the numbers will be different. So I can find it interesting what happens when they are listed on the stock exchange with their value. Will it stay there? Or a bit of speculation, or a high rating.

Mm.

Mm. Mm. Mm. Mm.

That's the question.

There are many interesting questions that will be answered, I think, at least when they...

Is there an AI bubble financially? Are we in the dot-com boom from the millennium shift? The difficult thing about forecasting for this, even if it's very high-value, is potential in that they might have built AI that can replace all human labor. And it's human labor that generates the economy and BNP and such today. Then the question is: Will we be as rich in the future if we do all the work? There will be other longer horizons, but in the near future, yes.

It's not very strange that a company that can potentially take over all the economy in the world is valued at $1 trillion. The world economy is in a state of mind, but it must be much higher. With the fact that there are so many already existing American companies valued at a trillion dollars or more. Apple has been there for a while, Google has been there for a while, and Microsoft is also over one trillion dollars.

The It's a bit exciting. And there's a lot of things here that you also think affects when they release and where they release and how they talk about things. And one thing that came was that all the big American AI companies, maybe not Elon Musk, but it was at least Google, Microsoft, Antropic and OpenAI, I think, that made a common public letter to, directed a little more towards politics in the USA or the whole world, I think.

We need to start to regulate more things. Not just regulate AI, for example. Opik had a statement about that. But the other ones were more like: "Our models are now at a level where they can do customized viruses, chemical war... They are so deep in the biological understanding that it's starting to get dangerous. And even if we build our own guardrails, we think you should monitor all purchases of substances and things you need for this." It's also exciting.

I And of course there is substance behind that.

It's a bit of a show to try to be morally good, but we've built this and we're also releasing these models so that everyone can use them.

It's like huge PR campaigns that just keep going.

I still have a hard time not... look at it with a bit of critical eyes or be a bit Questioning of some things, you can tell from the used models, they are veryThere is some kind of own reasoning in these models.

Hmm.

The subject works as it is in my opinion, or as we who listen to it.

Hmm.

So there is a reason why you can think about things. But the other thing that has pointed out is that this is still... Some things that can be a brake block. We talked to someone about it. It has no consequence thinking. It doesn't have feelings and other things to be really smart. And that's probably what makes it dangerous. Because it just goes action on action on action and can do things. Especially when we do a lot of new tools. To go around and get a lot of things.

Matt, What I'm thinking is that if it's just If we play with the idea that it's just a computer that has come up with a answer and then made an action on it, then we need to be somewhere at the control point. Because I don't think there will be any... I don't have a good or bad agenda behind me. It will be a person's responsibility to stand for. But what I have seen quite a lot of now with... A little before these great models came out, one started to think about how to replicate this Maybe some kind of consequence thinking or learning from experience.

Yes, we can create a lot of clusters of Yeah. in other programs, clusters, so it's like a second brain to AI to learn it. People are trying to test this a lot. Peter Steinbergen with Open Claw has obviously I'm trying to integrate a video somehow. I have seen some agents platform. They definitely tested different things.

Hmm.

The question is if we even need to get to or at least my vision of an AGI, that it actually feels and understands in a way that humans do, or can we do that with techniques of We have a choice, where we have given them "this has been bad, but before this it was good" and that they constantly read through these experiences. in that wayThey can't be dangerous if you use them enough.

Hi.

Yeah.

Oh. -digits and other values.

They had a Claude 4.7 Then came 4.8 very quickly. Then, I don't know what it was, a week later came Fable 5. No. Do you know, I haven't read much about it, but this Fable 5 Is it a myth or is it aYeah.

Hmm.

No, I understood it as if...

I've also been thinking about it and haven't been able to dig into it. I interpreted their release as "Now we're releasing Fable 5 and Mythos" Either Fable 5 is a Mythos-variant that is bundled in that direction or it will also be Mythos.

I interpreted it as if they would have several T's.

They have their haiku, sonnet and opus since earlier. And then it would be Fable Mythos on top of them as two different models.

ButHmm. I'm unclear.

I interpret Fable as the little brother to Mythos. It's not Mythos that is on-branded, it's a different model. And then there's also Mythos. Or the myth of the feeling. But that... It's hard to keep up with all the friends, but Fable seems to be extremely good at, above all, agent-based coding. I haven't tested it, but you have.

I have tested it a lot and there is definitely a difference.

Both the fact that it is very potent and very good, Opus 4.8 also took well, but it did things that Opus didn't think of themselves, which I probably wouldn't have...

Hmm.

I told it that I wanted it to make things more responsive or scale things, or I thought about different things, build tools, start up, I don't know, now there are only 73 agents at the same time, I don't know what all the 73 agents are working on, but it is like that anyway.

Hmm.

But I have to say that what I react most to is that The token cost, the cost, What I would like to credit or just use is Sky High. If you have started with 4.8 or 4.7 you can just sit and prompt and prompt until it runs out. If you run Fable it can run for two hours and then your use is over. So that's a huge difference. And I also think that something you can take with you if you have to sit in, is to program something or use it.

You need to use Fable 5, because 4.8 will make it much easier. Okay.

Yeah, well, that, yeah.

I've just switched to Fable for administrative workflows. And You don't notice the difference there, because that's not where the extra intelligence is needed. Opus 4.8 is brilliant in a lot of things. It's also GPT 5.5 and Gemini 3.5 Flash is also great. All top models are performing great in what most people do and want to do for them. But when it comes to these... The code project gives a rough description, it asks the right questions, it spins the right tasks, and so on.

That's where the development is happening right now. That's where you can see the biggest difference. I'm building quite a lot with Gemini. They haven't released their 3.5 Pro yet. They promised to release it at the end of June, if it hasn't been released in the last few days. It should be released soon. The question is if it will be released at the same level.

But 3.5 Flash is still a bit... It can work for quite a while and it can spin its agents and do things, but It doesn't understand the whole system. It doesn't do it right from the start. You still have to handle it quite a lot.

That's the fable now. Probably 3.5 Pro. It's about that there's a new GPT, if it's going to be 6 or 5.6 or whatever they have in the box. that we're getting to those models now. And Anthropic usually comes first. So that Fable came now means that the others will be throwing something out soon too. And that affects a lot, because we will come to the level of making a one-line prompt, like "I need this system", and then it's all back to the console.

Hello.

"Do you need this?" and it answers questions and builds implementation plans. You just need to be the orderer, you just need to say what you want. Because that's not the case, most people haven't been able to start coding, because you have to have both You have to stand up with a little frustration, you have to have patience, you have to know a little what you want to do. You have to, like, sanity check a little architecture and such things. And now we have come to the level that you may not need it. And there are more people coming.

I'm starting to get tired of the word vibe-coda, so I'd rather not use it. But that's what everyone says. For vibe-coda... your AI agent code team and you become a really good product owner. That's where everyone needs to be. Agile methods, understanding what the product should do, making a sanity check on whether this really works for me or our business or our customer or whoever we're building it for.

We are good at it. I feel like I still need a little technical competence to get it right.

Oh.

But we agreed at a meeting this morning that we both have built the same system. But I built it yesterday evening with the kids.

I built a kind of... To get experience on doing homework and cashing out at screen time. We've talked about it for a while and we've had some solutions in the vicinity of it. But it wasn't satisfactory to get a check in and get it approved.

So I wanted to systematize it a bit. And that one-shot Gemini 3.5 Flash when I ran it in Anti-Gravity. -Maybe 7-8 rows, but it wasn't meant for all the consequences.

I've been testing this for a while now, from the start you could have One shot a simple calculator.

And that's where the new models started to come to.

Hmm.

You one shoted tic tac toe the other day.

Hmm. Hmm?

It was really good. You can see clear progression. It's more and more complex things that it understands.

Hmm.

Mm-hmm.

What is this type of program? What do other programs look like?

Hmm.

What can we do? All kinds of techniques. A lot of things happen there too.

But as you say, if you're going to build something, Stir. Then you might need to have some understanding of some technical things. But it can be done in a lot of different ways.

Hmm.

Complex things, yes, that takes a bit of Amen.

Hmm.

Yep, yep, yep.

Tickles.

--------employees and developers who know these things.

They are on the job market now, because they have become more unemployed. So it's a good strategy, but ordinary businesses that have not had an IT department before-they need to start getting this competence in-house, I think.

Or find consulting companies that you can work with very closely so that you get a good flow in it. Hmm. -And have someone who can help you think about the right questions.

-Yeah.

You are the new me.

We can say this without being too obvious, that both you and I have a free half-hour consultation if you would be interested in discussing these issues with us.

If you go to AI now, it will think about a lot of things, but it will not think about everything. You also have to know what you should ask it to think about. And that is a challenging thing.

You can go to our website and book a time with us.

Yep. Hmm? Hmm? Yes, exactly.

klok.se and then down at the start page there is a guide. You can book directly in our calendar.

It's a great start if you just want to play a little. developer conference. umSo...

Hmm.

Oh. Hmm.

I remember two years ago when they released this, AI could take actions in your phone and do things and nothing would happen.

um Mm-hmm. Right.

What does it mean to use Siri?

Oh, there we go.

The few times I've tried Siri before it was more like a Google Home. I'm pretty slow to understand what you mean.

The first time we started any kind of AI assistant for a phone at all was Siri.

I don't know, 15, 18 years.

It's been a long time. I don't know exactly, but Siri has been with me almost as long as iPhones have been with me. In the beginning, it was really cool. Because then it could go in and look at what was in my calendar. It could understand the voice command, pretty basic stuff. But it has never managed a conversation. So it's one prompt and one answer. And the new thing now is that they really need a language model, an LLM, to build that bit. And they need a way to handle context.

So that they know who they're talking to and what they're doing.

Hmm?

Hmm.

And then they need tools to work in the apps. And all of that exists now. And then the thing is that they said this then, but then they said that we will roll out this in the near future. And there was no developer beta. Now there is a developer beta that people have downloaded at home and tested already on Monday. And shown that yes, it actually works. So it really learns to come to the fall. And if you're really fucking hungry for this, then you can join Apple's developer program, Public beta In July, I think, you can take some risks.

Hmm.

You can join the beta program and get tested before. I sometimes jump on them if it feels relevant. But otherwise, you can just wait. It usually happens on September 7th, when the OSs are released. That's when they release the new phones. When new phones come on the market, they will be updated with the new operating system. -I've had 50,000 pictures in my Apple camera roll.

Oh.

Good night.

I've had iPhones for a long time. Everything just stays there. Everything is indexed. It's already categorized by all people. I can search for a person and find all the pictures of them and all the places. Then it has the email and the SMS. I have SMS conversations with history 15 years ago. If it can do things, it will be incredibly useful. I don't want to tank in to the other services because I don't trust them like I do with Apple.

I would never dare to do it.

-It will be a big one, because iPhone is the biggest mobile platform in the world.

Hmm.

There are more Android phones, but then it's Samsung's version of Android, Google's version of Android, and everyone has their own ideas. But Apple has the same ecosystem everywhere, a lot of people use it. And that will probably make it possible for more people to get started, to see this type of flow. Because when you go to a KBT store, they know very little about you. Over time, they build up a little memory and such.

But imagine if they just have all the memory. It will be an incredibly usable system.

It would be very exciting if they could give these Google said the same thing a couple of weeks ago in their event.

I see now, I will be able to see the same thing there.

Ah, ah. But when that's done, it will be interesting.

But that also leads me to the idea that all these AI serviceswith a common man sitter or a common woman, right? who sits and triesIt's being pushed into places we don't know where to go from. Um. And as we have said before, what I think is worth trying out is these Codex, and not just chat GPT, because there is so much more potential you can do.

Hmm.

Mm-hmm.

I think it's fun to automate flows and other things, so I don't have to manually work.

But open them up and test them, Codex or Cloud Code, Co-work for some part. Um. So, um, But you also mentioned I will be quick with the cars, but I see in the news article that we have written about Nvidia's deep collaboration with Hyundai and other companies.

Hmm.

And there is a lot in robotics and other things. We have talked about robots and other things before, but also in cars. Cars are just getting smarter and smarter.

Hmm.

There are cameras on the right and left. Autoprompter, and it can easily recognize people and read a lot of things. There are signs... You name it.

Hmm.

I have been working with cars for many years and I have always been convinced that these systems are a great help.

Hi.

You should never trust them.

Oh, he's... Oi, oi.

I see it the same way as with AI, but Google and all those who I think they had to take a feeling that they see it the same way. If you google it, you will get an A-generated answer. And it doesn't have to be correct. Even if it looks like it came directly from the web, it can be a wrong answer.

All right.

There was a case in In Germany, the AI overview service is called. where they got banners and the court decided that this wasn't okay because it was a very wrong answer.

No.

I don't think that the trial is complete yet, but... AI is rolled out as something that I feel like you should be able to rely on. And you can certainly do that in some cases, but when AI comes into robots or cars where we have previously seen these systems as just aids I see that it can be a bit problematic that people just think that if I have an AE car So now it can do what it wants and I'm responsible for something.

Yeah.

It's self-sufficient, I don't even have a steering wheel anymore. Um...

Bye.

Hmm. Hmm.

Hmm.

Hmm. Well then, as a It was a boom two years ago, the self-proclaimed cars and the debate. It's been quiet for a while now, it will probably wake up again when it really starts to work. I saw an insert in the debate that compared it to how much... -The debate before was if the AI in the car should choose between the driver who dies or the one who goes over the road who dies, who makes the moral decision and so on.

Oh.

They were quite hypothetical. Now it's just like, or it's not going to die. -I feel quite uncomfortable when I drive with people I don't think drive well.

Hmm.

- It will be exciting, Skåne will be outside of AI, but AI will become part of society.

I don't think I feel that with an AI. I don't expect to drive well, so I will probably be more relaxed in an AI car than in a human driver that I don't trust. When you say that they are wrong on the road or something. That's my kind of technical optimism. I think we're going to have to think about all the security around it. Do we trust that all cars are connected? Do we trust all suppliers? There will probably be some car manufacturer who just rushes out their autonomous cars a little too fast.

Um, And then they get hacked. And then the others who are safe come along. And then we get that separation, just like which system is safest, which car you dare to buy, and it's probably the most expensive. So it's probably ten years of challenges like that. manand stuff like that, they don't all have to start at the same time and so on, how much a car can actually be used. And there was a little, I tried to draw out finished research on this, and then there was quite a lot of calculated evidence that we might need about a tenth of the car fleet today. So the world can take away 90% of all cars and people can still drive with, in a large sense, personal transport and maybe a little co-transport.

And that becomes a huge positive environmental impact and it's not like, the effect has not been that everyone gets their own car.

-Yes, it's completely different conditions. And we talk so little about that. We talk so much about who the car should choose to kill.

So we have to take the whole picture, I think.

Mm-hmm.

If we can cut We're going to take a lot of our cars that we have smaller, and we're going to pick up people Um, Who's going to own those cars?

Yeah.

Hmm?

Hmm? That's a great question.

The...

Hmm?

I remember the name of the agenda, 2030, that you talked about. And that there was a slogan that said: "You will own nothing and be happy" or something like that.

Hmm? No, no. Hmm.

That the state will own You will rent things. And yes, it might be like that. I don't know how it will be, but it might have been more efficient if we had had it.

Hmm?

I just hop on the car that drives past me here, because it's another one that will get a job. About the same place at the same time. Now we're talking about environmental protection.

Hmm? I'm going to go with the LDR. -financial advice ...consumption and such.

Mm.

And then a trip locally to a city would probably just cost maybe 8-10 kronor with such an autonomous car. And that's a lot cheaper than those 40 kronor. So as soon as that's possible, I think people will use that more than they will use buses. And then collective traffic disappears. But then it becomes private owned collective traffic. And we take that as a bridge over to... -Okay. That's something that will be a big issue in the next term.

Mm.

I think people need to factor in what they vote for in the fall. There are other issues as well. I don't think you should only vote for the EU. But since there won't be a big political debate about this before the election, we have to choose representatives who navigate these issues. We'll have to have autonomous vehicles during the next term. It's European-based decisions. When it's done in one country, it'll be done in all other countries. or need to be passed through.

And that makes it Isince long ago.

Yes.

I've been collecting data since 2018, but that's eight years ago. So there's a lot of data. So go and look at it if you think it's exciting. Because everyone is going to the election halls. It was 100 days last Friday, so it must be 93 days today. It's in that feeling too, that many days are left. So you get a little stress factor to read about. But I think it's important. I don't want to affect anyone where they vote.

I don't know what the right answer is either, but I built it because I myself-Now We'll have to rewind to what we had last week, with testing some of these agents.

Oh, never mind.

Yeah.

Of course people want to hear about the practical stuff.

I don't have a very good practical tip. I don't know if you have anything. Have you tried anything?

You've already said that today. If you have ChatGPT, you can download Codex and test-lab a bit in it. If you have Claude, you can download Cowork and test-lab a bit in it. If you run Gemini, you can download Antigravity and test-lab a bit in it. Grock vt17, but I don't think anyone has a Grock generation. I don't know if Grock has an agent like that. I haven't labbed there. But the three big ones have this.

And if you go into Copilot, you don't have an app for it. But they have said during their build conference that this type of agent-like situation will come to Copilot. And where it ends, whether it's in the Copilot app or elsewhere, we don't know yet, I would say. But in the other three, JetGPT and Google with Gemini and Anthropic with ...models, there are specific apps.

Yes, it's possible.

Good. Yeah, I pay for it. I have max...

Download them, test them, even if it's just on Word files. You don't need to write code. It's just the feeling of working in another type of app with the same type of agents or AI makes a huge difference. And if you have access to Cloud, you can test Fable without pure token costs. It was something like the date they released it.

You get some kind of trial runs on it, but it's limited for a while.

All right. Sorry, I gotcha. Yeah. Yeah. Hmm. Hmm.

The other thing I've seen, that they've put some time on, is that As I understand it, you need a max account after the 22nd of June to have access to Fable. So this feels like a... A bit like how Notion did the Genshin Impact. It's free until this point and then you try to trick people and then you have to pay a higher fee. But until the 22nd of June you should be able to use Fable. As I understand it.

And then free roll-down. We'll be there in half an hour.

Oh, yeah. that Hmm.

And next Friday it's midsummer, so we won't be live, but the day after that we'll be back, because we're not on vacation yet. So it'll be one broadcast for this summer, maybe two. We'll see. So får ni jättegärna höra av er. Skriv kommentarer, höra av er. Vi pratar gärna om det ni vill att vi pratar om. Så skicka under veckan, hitta på saker. Vi försöker gärna fördjupa oss i det som intresserar de som lyssnar.

Okay. Oh.

Exactly. We'll post it in the comments as well, so you can all link it. Thank you. Have a nice weekend.