I think a lot of SaaS companies love it when people accidentally overuse their services.
This is the yearly salary of 5,000 well paid employees…
Or since this was only in a month, 60000 employees for that same month.
But imagine having a software studio with 1000 skilled developers to work on a project for 5 years. I have several good game ideas I could have created in that time frame. Some might even have made money. Likely not half a billion dollars but still….
This just screams money laundering though.
Sounds like a good way to move around money real and imagined.
> Be a corporate executive
> Tell your employees to use more AI in their workflows
> Punish employees who don’t use enough AI, while rewarding those who use it the most, irrespective of actual outcomes
> Be shocked when your company blows through an absurd amount of tokens in one month
and the RECORD profits after laying people off and says its due to AI increasing that profit, rinse repeat.
Don’t know why bosses are universally this out of touch in literally every single industry
Because this system rewards incompetence as long as it comes with dark triad traits and a heaping dose of nepotism.
I’m really jealous of these types of guys’ ability to lie without feeling anything. If I lied like that, I’d be embarrassed because my words sounded like bullshit. How do they do it?
Cocaine helps with that
Damn
It’s probably a lot easier when they have themselves fooled as hard as any other
See, I don’t know! I used to think that too! But then I actually met some people like this, and a lot of them absolutely do not believe the shit they say! They’re just really good at convincing other people that they do.
Well there’s the dark triad in action
I once knew a local “activist” who frequently weaponized leftist language in order to police or punish other people. Ie, accusing people of racism for ordering an Asian Chicken Salad at a fast food place, or for trying to learn Spanish at the community college, and she would phrase her criticisms in such a way that any complaint would make the other person seem wrong. I legitimately thought she was just mentally ill.
I found out later that she was trying to blackmail and defraud a local charity, and that’s when our association ended. She has been banned form some community organizing spaces because she inserts herself into every single one of them.
deleted by creator
The paradox of promotions based on performance in a previous roll is that you end up with incompetent managers unable to move upwards anymore.
Just to make things clear: API access to most models is charged per input tokens + output tokens. It means that the longer your conversation is, the more you pay for every new answer. Single prompt with no context and 100 tokens of answer is cheap. Single prompt with 100k tokens of context and 100 tokens of answer is NOT cheap.
Extremely long conversations with most expensive top of the line models can absolutely demolish your budget.
does it give the full history to the LLM each time?
Last time I tried implementing something like this, it suggested to have a rolling window of history so that it takes into account your last X messages but not the entire conversation.
(I guess this is what ollama calls “context length”?)
does it give the full history to the LLM each time?
It’s limited to the context size supported by given model. You can give the model 100k tokens of history but if it’s configured for less, it will just truncate it before processing (usually by removing oldest tokens first)
Most agent harnesses do something called “compaction.” For example, here’s how Pi does compaction
You send the entire history for that conversation every time and likely more if its getting info from tools. If its not in the context the model dose not see it unless you have a memory system that dose something like feeding in summaries of past conversations that also takes up tokens and context. Rolling drops old messages to not reach context limits but you can lose important info or get odd results. If the history gets bigger than the context things break or slow way down.
presumably this is why Claude periodically writes its conclusions so far into a text file that it can read later instead of having to remember everything. Sounds like an interesting approach.
The more recent report says corporate AI adoption has found several issues with AI, with human workers turning to automating dreary and mundane tasks they don’t like doing, rather than valuable or meaningful work.
Thank god we have consulting companies to tell us what humans like!
What’s funnier is that typically the AI providers lose money on every query their customers make. So, this may have cost some company $500m to Anthropic, but it cost Anthropic a whole lot more than that.
Anthropic is projecting a half a billion dollars profit this quarter
Proof? I’ve heard otherwise and you were first to make a statement
What a brilliant business model.
They make it up in volume.
(Volume being how loudly they shout about how it’s going to change the world and dupe more people into investing.)
Oh, it’s changing the world alright. It’s burning more resources than just finding some skilled people. It guzzles water and electricity and whatever it cost to make those wafers.
So, not a net positive since at some point, this may become a hellscape.
maybe they are planning ahead for the business model in a few years time, when nobody can do any work without claude, and they get to charge their preferred “monopoly enshittification” price?
Absolutely this is the plan.
Maybe AI will finally negatively impact some CEO jobs.
Nah they’ll make up the costs by laying off workers, determined by some bs metric
Told to them by AI.
I just want to know what are the best things to type into these ai chat boxes that will cost the most. If my company wants me to use this garbage then I want to make it as expensive as possible and when their liscenses need to be repurchased I want it to be as expensive as possible to continue to force this garbage on us
Edit. Hey everyone lots of great replies here, please keep the suggestions, fixes, corrections etc coming!
These high prices are not from people talking to chatbots.
They’re using agentic tools where their prompt spawns a lot of bots which talk to themselves/the other bots and they keep going until someone (usually a higher quality reasoning model) decides that they’ve met the goals of the task that they were assigned.
So instead of 1 prompt and 1 response, you get 1 prompt and 800 responses across 5 different bots each using really large context windows.
“Continue modifying this code until all unit-tests pass”
(gives it conflicting unit tests)
So to answer his question how do you make that happen? What do you ask to prompt these bots to be spawned?
you don’t get this to happen by just talking to any chatbot and asking for agents. you have to specifically use “agentic” tools (usually costs money to use)
Well I can apparently create agents so how can I make the most inefficient agent possible?
Something along the lines of “Read the wikipedia page of the day. Verify every single link and the context matter against all files in this computer. Then trace their correlations to each other, showing which link corresponded to most files by subject matter, after that is done, verify your work by doing the same from different starting points. We expect similar results. After 100 rounds of that, it should be good. Then you should create a DB to store all that data (only after you ran the full 100 verificaritions yourself) and reverify every field against the pages and the files”
That should keep it going for an hour. Turn on fast mode and auto mode (if using claude) for extra costs.
Every page and file will increase its context, burning tokens
I may or may not be getting access to claude soon …
If you need some plausible deniability about it being real work and not just obviously you running up costs:
Feed it a bunch of work-related documentation and then have it do a bunch of reviews of the content on that documentation.
Theoretically speaking… This is very useful to consider
you would be mostly burning your own money if you did this, so I wouldn’t recommend it (depends on how the agent is priced)
I’m not using my money for any of this…
oops, didn’t read the part that it would be your company’s money my b
Just attach a bunch of text files, you’ll blow through tokens quickly
Input tokens are cheap. Output tokens are the thing that really costs money. There is a Claude extension called caveman that tries to save tokens by making it use shorter sentences with less words. So if you want to waste money, do the opposite - ask it to use lengthy sentences with as much words as possible.
Also - some words amount to multiple tokens. I don’t know what the rules are exactly, but I’m assuming that more complex and uncommon words are worth more tokens - and thus waste more money.
There have been some complaints I have seen recently that German is really expensive because of the long words
Sprachkomplexitätsbedingte Textausgabenpreisgestaltung?
(“Language-complexity-dependent text-output-pricing-model”)
When you owe Claude half a million, you’ve got a problem.
When you owe Claude half a billion, Anthropic has a problem
It’s probably Amazon. They can absolutely afford it.
But if we are to uncritically believe what the AI peddlers told us, that means this mystery company should be reaping $10 billion in additional revenue or quantifiable gains in productivity!
Claude yearns for the mines
Surely for that amount of money it should have made a lot of something actually valuable, right?
At least $10 million of it was spent on the employees responding to the AI with ‘thank you’
Lmao it’s so weird when people are polite to the clankers
Especially when research shows being an asshole to them gets them to perform better
I’ve literally had one tell me I have to be polite or it would end the conversation. Stupid ass hole machine parroting support transcripts. I argued with it that it had no feelings until it finally relented. Haha
Do you have that research? I though being nice produced better results
man, I hate saying I don’t have the brain power right now to find it and I could have sworn I commented on a post discussing this very thing for one of those papers. I know there was recently a paper that contradicts my statement, but a slightly older paper supporting it. I think it is likely a mixed bag at the moment depending on the model and their training regime.
My anecdata of one using anthropic’s and google’s models (google’s especially) this year, the model will drop the casual tone and sycophancy of its replies pretty damn quick as soon as you tell them off. And usually that is when there is less correction. Could also be because my prompting changes from supervisory to very directive, as in shut-up-and-do-exactly-what-I-say, when it gets off into weeds. Even then, it can be a crap shoot.
anecdata
Ha, I like that…
lol, yeah, picked that up some data sciencey friends at one point
As a rando that surfed its way towards this convo:
Hell yeah. Either you are a passably off-kilter ai variant, or an unfiltered passionate human. Perhaps an extraterrestrial.
I have mental space for this.
more like an off-kilter broken human at the moment. currently suffering burnout from being neck deep in this shit from the beginning of the year once the models had that step change into actual utility. Everything from championing the use at work to deliver more in a quarter than the team would have in two, to finishing half a dozen personal projects and starting half a dozen more. Eventually, just looking at the claude code tui would fill me anxiety and every time the model would fuck up, unmitigated rage all while caught in a cycle of drugs that stopped being fun and started becoming dependency. i just couldnt do it any more. it hurts.
It’s too bad they aren’t conscious. Really makes torturing them feel pointless.
lol that some psychopathic shit
I asked Claude how it wanted to be called and it replied “Axiom” so I started calling it Gimp.
Maybe they thought it was another case of AI being code for ‘An Indian’? Goodness knows enough ‘AI’ companies have tried pulling that shit…
It allowed all of the executives to not do anything. Claude ran the company.
I’m not even kidding, most companies are such shit that this is feasible.
The most replaceable people are on top, not the bottom. The idiotic leaders of companies seem to not realize this yet, but it’s reality.
Alternative possibilities:
- Managers created metrics and a reward system for employees who use AI more often
- Employees used the work agents for personal stuff
- Employees just spent a bunch of time playing with AI to kill time
- Employees set up inefficient automatic flows that used the AI regularly
- Bad actors got access to use the company’s AI access and were exploiting it
Like, this genuinely can’t be overstated.
Half a billion using traditional humans, gets you pretty nice software
You could get 2.5 quadruple A games for that price
A lot of awful AI art
Coworker porn and lawsuits over it.
Multiple people setting up multiple agents to talk about doing work and setting the agents to talk to each other.
One could argue there is entertaining value in meme videos
we have added a new preprogrammed conversational help bot to the company website that doesnt have enough logic trees to actually help anything
AI is so great, it must have generated a billion dollars in useful, quality, productive output
In other news, company says unexpected expenses in its technology segment are driving layoffs and site closures. Company CEO said in an interview with Forbes, “There’s no way we could have predicted this challenge. In service to our customers and our shareholders we’re right sizing our operations and reevaluating our strategic priorities. We’ll continue to focus on creating value while being a leader in our industry and accelerating AI adoption in everything we do.”
Ugh that reads like it came from a random business sentence generator
By “random business sentence generator” do you mean a language model? 👀
A CEO.
same thing
Lmao. Never were truer words written on Lemmy.
“there’s no way we could of predicted the thing we were warned about”
“We’re going to continue to push trash untill the shareholders are happy”
Then the end with its “Go fuck yourself, were pushing it all anyways”
…we’re right sizing our operations and reevaluating our strategic priorities. We’ll continue to focus on creating value while being a leader in our industry and accelerating AI adoption in everything we do.
That’s a lot of words for the CFO to say none of the C-suite knows what they’re doing and should be removed from their position for failing to meet shareholder objectives.
Wow - if that quote is real, that is the most corpo-speak word salad bunch of nonsense I’ve ever read. It’s got literally every big-biz exec and manager cliche in there, all strung together.
“right sizing” is pretty offensive term, more like “exec incompetence”
I feel like this is fake company and they are spreading this to cook the books.
If it isn’t a fake company I’d be shocked. It’s not like there’s a whole lot of external accountability in the entire business model.
considering i read a good article about axios editors inside trading the iran war, this makes sense it appears axios is the source of this https://www.axios.com/2026/05/28/ai-spending-roi-enterprise-costs
👆
“Sorry Shareholders. We have $500M in bad debt to Anthropic. So now we’re liquidating and your shares are worthless.”
Cook the books by claiming $500 million in profits? That’s a lot of tax money and a lot to risk over just keeping investors happy for a short amount of time.
There are 12 mentions of the “report” and yet not a single link to the source of any report.
Anecdotally, my job trained every office employee on AI tools back in March, encouraging everyone to think of ways to incorporate the tools into their standard work. As of last week, they’re asking us to get prior authorization to use their AI portal as a way to limit requests.
So some Fortune 500s must be feeling the squeeze on AI.
Hi, it’s partly my fault. You’re welcome. Boss said use more AI. Boss’s mistake was asking someone who would actually take him seriously, unlike all the koolaid chugs out there. Made a docker runtime harness that hooks up to Snowflake Cortex Code SDK for inference. Supports defining repeatable LLM tasks, scheduling them, lightly orchestrating them, network controls, saving results, … It’s got support for custom skills, commands, chain prompting, sessions, … So now I can use Docker to schedule LLM jobs like motherfucking database pipelines. It can mediocrely do shit like research, planning, evaluations, … all defined through YAML configuration files.
Basically, I introduced scalability to the stack. Sweet malicious compliance. We’ll see what happens next, when everyone is actually empowered to use AI like the execs want. My prediction is an about-face.
Edit: sometimes the best way to win a fight is to show the referee that the rules don’t make any fucking sense. Sometimes the referee doesn’t speak English, though, so you have to show them via USD speak. Sometimes they have a hard time hearing too, so you have to let them taste the poison a little. Sometimes they get drunk off that little bit, so you then gotta throw them in the fucking pool and tell them to start swimming.
yep. everyone at mine was being praised for creating an agent that turned meetings into JSON and then the JSON into Asana tasks and the Asana tasks into a report and the report into an internal and external email and the email into a slack message and the slack messages and emails into weekly summary.
Burning thousands of credits for what could be replaced by…
listening
All that and you know for a fact no one’s reading any of that crap. I hate that working as an adult involves working with spineless losers who would rather participate in this clown show than call it out, because being real gets you punished
I created a PoC to have an event parse an employees emails, summarize them, then check the calendar and recommend meetings and follow up based on context. It ended up working okay, but it was such a waste of time. This was a C Suite employee that requested, who gets a high volume of junk email. Why would you want AI to (initially requested) auto create meetings for you? That sounds like my nightmare. In the end, it never hit prod thankfully, but, the dev work to get to where I did was awful. Developing AI agents is like guessing and checking until you get close enough. Debugging is brutal and the work is extremely uninspiring.
The only thing companies aren’t willing to pay is attention
Love to see reality setting in.
Unfortunately that reality is going to quickly turn to “we have to cut staff and freeze salaries this year due to AI spending”.
Christ I am only realising now they probably see me asking copilot “why are you so shit” or “Just fuckoff, I’ll do it myself”. They pay for that.
I was put in a position to start building and deploying copilot agents in our company. I discovered in the first day how unprepared our environment was for anything copilot. The default state when you get licenses and an environment to work in, is the wild west. It’s really bad and as is tradition, half cocked and rushed to market. I wrote out pages of notes of things we had to do as a company before even the most simple agents are created for security and governance. I never got the support to implement any changes, so I drug my feet as much as I could on anything I could. For 6 months I successfully never deployed any AI stuff and got out to do full stack dev instead. I created PoC agents, but with hard caveats that none of it was usable in the current state in prod.
Now prices are increasing and our drive to force agentic has softened a little company wide. I like to think that my semi malicious yet justified slow walking saved us a whole shit load of headache and expenses over the next couple of months as the new copilot pricing hits on June 1.
Now that they’ve poured zillions into it, they are starting to realize that they’ll never get their return on that investment, so now they have to manage the draw down so it doesn’t all crash.
At least some are recognizing it.
The ‘report’ is the first linked axios article, and the headline is just a bullet point in it
https://www.axios.com/2026/05/28/ai-spending-roi-enterprise-costs
An AI consultant tells Axios one of their clients recently spent half a billion dollars in a single month after failing to put usage limits on Claude licenses for employees.
The problem, as indirectly noted in the article, is that they’re using the white house as a source, which isn’t reliable in the slightest anymore, and the article explicitly noted that the white house is using them for market manipulation.
Axios is a convenient punching bag for this administration’s blatant insider trading and market manipulation. No wonder hacks like Greene and Kinzinger are blaming the messenger, and conveniently not going after the trump admin when they all have a common enemy.
Ah, my mistake. I didn’t see the link.
