Artificial Intelligence

[Spekulatius]

3 hours ago, gfp said:

 

I don't know the answer to that.  I currently pay $22 per month to OpenAI, I paid $99 for a year of Gemini (limited time offer), and I paid something like $220 for year of Anthropic's Claude "Pro"

 

I think Anthropic is the most disciplined about not giving away token usage for a big loss.

 

I'm not investing in any of the model companies and it is possible that they end up a little like the Ford CEO shaking in his boots at the thought of Chinese cars entering the US auto market.

 

So far, Anthropic has done very well finding paying customers and I personally can't imagine going back to lower tier or free versions of these subscriptions.  These are a phenomenal value.

 

 

I assume you use them as a consumer - what are your use cases for the pro  AIversion. I use notebookXM, Gemini and so far have not really exceeded what the free version give to me.

 

I would love to use the Claude or a Gemini Pro at work but can’t because we are limited to Copilot which I have found mostly useless. I think Gemini on my phone is better than the paid for Copilot at work on my laptop.

[nsx5200]

AI is currently in the promotional phase, so a lot of it is given away for free, if not below cost.  Just like how cloud/photo storage companies were giving away spaces with few conditions attached during the market share grab phase.  After a while, the service providers will shift into profit/cost optimization, and free suddenly comes with a lot more conditions that steers people towards paid solutions.

 

The fact that many of these AI/token providers have already started imposing some limits already, even on their paid subscriptions, portends to some serious cost issues.  The primary lever for determining the eventual winner(s) of this AI war may be cost.

[Peregrine]

Using AI to write a white paper about AI adoption...what could go wrong

 

https://www.ft.com/content/b3828e92-4961-4b39-84f0-c42f33be3c3f?syn-25a6b1a6=1

[bargainman]

On 6/11/2026 at 8:03 AM, Libs said:

 

gfp,

 

How much would you be willing to pay? I'm trying to figure out if users will ever cough up enough to cover Anthropic's (et al) costs....

 

I'm kind of curious if anyone here is a developer using AI?  Or in another field where AI, especially agents are making a huge push?

 

Regarding "How much would you be willing to pay?", that's a reasonable question.  But the ones to ask are, "How much does using this tool increase your productivity, and how much does that productivity cost/deliver in value?".  There's nuance there, but if you're a software dev making 100K annually, maybe driving 250k in revenue and using agents increase your production by 12%, is 1k a month a bad investment?  What if it 2x's your production?  Now, does that 2x in production result in 2x revenue?  probably not?  Uber CEO seems to think that not.

 

But it sure writes a lot of code.  To me so far the biggest boon has been in 0 to 1 projects.  I've created about 5(?) projects I would've never been able to even start.  Did they increase revenue?  Not all of them, plus I'm deeper down in the org, so there's no one-to-one.  The actual production code is a bit harder to quantify.  Teams still need to figure out how to get the massive amounts of code properly deployed, but that's a known bottleneck and will keep evolving.  i.e. the models are really just the baseline right now.  It's the agents and frameworks that are making the difference going forward...

[MungerWunger]

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

 

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

 

Access to all other Claude models is not affected.

 

https://t.co/bwn0sximKZ

[gfp]

25 minutes ago, MungerWunger said:

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

 

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

 

Access to all other Claude models is not affected.

 

https://t.co/bwn0sximKZ

 

It's too bad, I was liking Fable 5 this week.  My current favorite.  It's shut off for now

[Spekulatius]

2 hours ago, MungerWunger said:

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

 

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

 

Access to all other Claude models is not affected.

 

https://t.co/bwn0sximKZ

Payback for the DOD thing a few month back.

[Spekulatius]

LOL:

Seems like Anthropic IPO will need to wait a little longer. How do you account for the risk of AI regulation?

[sleepydragon]

This probably is good for AI supply chain? These country will need host their own data centers running their own models , which means more chips and chip equipments

[Spekulatius]

It will be interesting how the export controls on AI products will impact the valuation and revenue trajectory of the LLM companies OpenAI and Anthropic . To me, it looks unlikely that the export control will be limited to Anthrophic since other LLM’s have similar capabilities , if not now, with a small time lag.

[NnnnotSoSmart]

Planetary Intelligence: AI Leaves the Internet, Moves Into Space, Earth Gets a Nervous System

This episode is about the collision of Earth intelligence, orbital compute,

 

[Spekulatius]

I am old enough to remember Perplexity - RIP:

 

[frommi]

On 6/12/2026 at 1:10 AM, Spekulatius said:

I assume you use them as a consumer - what are your use cases for the pro  AIversion. I use notebookXM, Gemini and so far have not really exceeded what the free version give to me.

 

I would love to use the Claude or a Gemini Pro at work but can’t because we are limited to Copilot which I have found mostly useless. I think Gemini on my phone is better than the paid for Copilot at work on my laptop.

I am using Chatgpt Pro right now with GPT 5.5 on High thinking mode and i am very impressed by what it brings up. It can create a whole investment thesis in 4-5 minutes that has little holes in it. So yes i am paying. I also developed the prompts for it using the model. Probably an even better value proposition for 20€ per month than any substack subscriptions. The free gemini model is good, but the frontier models are even better

Lets see if that leads to better investment performance 1-3 years down the road.

I am also more and more impressed by what Copilot is doing inside Visual Studio (i think it was using opus 4.8), yesterday i had a problem in a 10 million line of code base and it fixed a unittest on its own by changing some files from a little prompt that sounded like a crying baby, something like: My unittest is not working, must be something with loading the modules, fix it please and 3 minutes later everything worked.

Edited Saturday at 10:42 AM by frommi

[Spekulatius]

Pretty good talk about the history of AI:

 

[frommi]

Just found that GLM 5.2 is now also hosted in europe, costs only half of ChatGPT, and i prefer its responses over GPT 5-5. Very interesting to see how GPT 5-6 is compared to that. GLM 5.2 oneshots some answers where i need to repeatedly have to query GPT5.5 something like "check it again looks wrong" to finally get it right.

https://lumo.proton.me/

[Sinbius]

1 hour ago, frommi said:

Just found that GLM 5.2 is now also hosted in europe, costs only half of ChatGPT, and i prefer its responses over GPT 5-5. Very interesting to see how GPT 5-6 is compared to that. GLM 5.2 oneshots some answers where i need to repeatedly have to query GPT5.5 something like "check it again looks wrong" to finally get it right.

https://lumo.proton.me/

 

80% of U.S. startup use Chinese AI...

Any data that doesn't make someone think that Chinese AI has already won and will continue to win vs U.S. AI?...

Edited yesterday at 12:07 PM by Sinbius

[rogermunibond]

Forward deployed engineer going to be a growth job

 

https://www.cnbc.com/2026/07/02/microsoft-commits-2point5-billion-6000-employees-ai-implementation-unit.html

[Spekulatius]

7 hours ago, rogermunibond said:

Forward deployed engineer going to be a growth job

 

https://www.cnbc.com/2026/07/02/microsoft-commits-2point5-billion-6000-employees-ai-implementation-unit.html

They used to be called application engineers or customer support specialist. Must be a new thing for these billion $ companies that such a thing exists.

Edited 12 hours ago by Spekulatius

[Longnose]

On 6/12/2026 at 5:57 PM, bargainman said:

I'm kind of curious if anyone here is a developer using AI?  Or in another field where AI, especially agents are making a huge push?

 

Im not a "professional developer" but a hobbiest and I pay personally $200 a month for the top tier of gemini ultra. I have vibecoded about 6 apps now from super simple to moderately complex. These are things I would never had the time or aptitude to develop on my own and wouldn't have been worth paying a developer to do. But now I can accomplish these projects while watching youtube and project managing the agents. 

 

I see no future in which I dont pay for strong AI models. 

 

Gemini Ultra comes with spark beta too. Which has bene super cool. Its kind of like having an executive assistant in my google workspaces. It actually drafts emails for me in my inbox. schedules stuff for me. I can task it with research or objectives and it does the tasks for me or creates documents in sheets/docs/slides for me.  Ive been rather impressed with its outputs and multistep reasoning vs just asking questions to gemini in chat. 

[rogermunibond]

@Longnose how do you keep track of all the tasks and assignments that Spark is working on?

[Longnose]

@rogermunibond I asked it to make a project tracker in google sheets and keep track of projects and status and inputs required and create a scheduled task to review each morning if any changes or new projects need to be added. Does require me to review regularly but helps me keep the rolling project list. 

 

I also, have several other scheduled things. It summarizes each weeks worth of emails each weeks and sends me a summary email of the past week with any open things i haven't responded to. I have terrible inbox management.  As well as a few other scheduled tasks related to projects i am working on where it updates the spreadsheets and or documents and sends me reminder emails to go review them and provide inputs.  

Edited 5 minutes ago by Longnose