> The noticeable spike [~20 percentage points] in May in the figure above [tool invocations] was largely attributable to one sizable account whose activity briefly lifted overall volumes.
The fact that one account can have such a noticeable effect on token usage is kind of insane. And also raises the question of how much token usage is coming from just one or five or ten sizeable accounts.
It is quite interesting to ponder these usage statistics, isn't it?
According to their charts they're at a throughput of something like 7T tok/week total now. At 1$/Mtok, that's 7M$ per week. Less than half a billion per year. How much is that compared to the total inference market? And yet again, their throughput went like 20x in one year, who knows what's to come...
Yes, but that token growth chart looks linear to me. There's the usual summer slump and then growth catches up once the autumn begins, but if you plot a line from the winter growth period at the start of 2025 you end up roughly in the right place except for an unusual spike in the most recent month (maybe another big user).
I'd have liked to see a chart of all tokens broken down by category rather than just percentages, but what this data seems to be saying is that growth isn't exponential, and is being dominated by growth in programming. A lot of the spending in AI is being driven by the assumption that it'll be used for everything everywhere. Perhaps it's just OpenRouter's user base, but if this data is representative then it implies AI adoption isn't growing all that fast outside of the tech industry (especially as "science" is nearly all AI related discussion).
This feels intuitively likely. I haven't seen many obvious signs of AI adoption around me once I leave the office. Microsoft has been struggling to sell its Copilot offerings to ordinary MS Office users, who apparently aren't that keen. The big wins are going to be existing apps and data pipelines calling out to AI, and it'll just take time to figure out what those use cases are and integrate them. Integrating even present-day AI into the long tail of non-tech industries is probably going to take decades.
Also odd: no category for students cheating on homework? I notice that "editing services" is a big chunk of the "academia" category. Probably most of that traffic goes direct to chatgpt.com and bypasses OpenRouter entirely.
Good points here, particularly the ends not justifying the means.
I'm curious for more thoughts on "will drive more and more people out of jobs”. Isn't this the same for most advances in technology (e.g., steam engine, computers s, automated toll plazas, etc.). In some ways, it's motivation for making progress; you get rid of mundane jobs. The dream is that you free those people to do something more meaningful, but I'm not going to be that blindly optimistic :) still, I feel like "it's going to take jobs" is the weakest of arguments here.
It happened before, and it was an issue back then as well.
Mundane job may be mundane (though note that it is sometimes subjective), but it earns someone bread and butter and it is always economic stress when the job is gone and many people have to retrain.
If we were to believe those of us who paint this technology as mind-bogglingly world-changing, that someone is now nearly everyone and unlike the previous time there is no list of jobs you could choose from (that would last longer than the time it takes to train).
If we were not to believe the hype, still: when those jobs got automated back then, people moved to jobs that are liable to be obsolete this time, except there is also just more people overall, so even purely in terms of numbers this seems to be a bigger event.
Many people would not have found this cleaner without Googling, reading reviews, etc. While that may not be an ad directly, it's part of the marketing budget. So what needs to change?
We've had markets for all sorts of domestic help for centuries before we had computers. Perhaps more relatable, think about how your parents might have found such help.
Before that, there were classified ads in papers. Those were lightly vetted by the local newspaper. Also, with a warrant, the police could generally track down the person that placed the ad, which broke a lot of bullshit scams. (Like house sitters that don’t exist, but are instead getting lists of people that will be out of town.)
That tells me that modern advertising isn’t making things more expensive, otherwise companies that spend money on it would be crushed by companies that stick with the old ways and can undercut them.
No, I can't do it that way anymore: my local paper doesn't have classified ads anymore. There are only different online versions, which are a lot cheaper and globally accessible, thus have a lot more fraud.
You get it. A couple phrases I live by (taught to me by the haggling parents generation); "you never know unless you ask" and "the worst they can say is, NO" These don't need to just apply to goods and services either. They have lead to very interesting and life altering experiences that wouldn't have happened if I didn't ask a one sentence question.
Yep. And sometimes a nearby independant contractor who advertises once or twice a week on Facebook or in the local newspaper is going to provide a better service experience than the one blasting TV commercials on the local channels.
The nearby contractor who gets all their work through referrals is by definition better than the one who needs to blast TV ads. The best people are basically never on the market.
Not really by definition necessarily. But yes, it does seem very likely that referrals are the stronger signal.
In some sparser places there might also only be a couple contractors working anyway. Might be able to get suggestions just by asking around wherever you get permits.
The problem is that Google operates both the buy side and the sell side of the monopoly-scale ad platform. They're the only party in the transaction who sees what both parties are willing to pay (imagine on eBay knowing what everyone's max bid was), and sets the algorithm to maximize their take from all parties.
They've already lost the case with this, and are currently trying to prevent what needs to change: Google must be forced to divest large portions of its ad business to reintroduce competition in the marketing space.
>They're the only party in the transaction who sees what both parties are willing to pay (imagine on eBay knowing what everyone's max bid was), and sets the algorithm to maximize their take from all parties.
Is there any evidence of them abusing that knowledge? Or was the lawsuit over them having a monopoly and/or anticompetitive practices?
I mean, the fact they abused that was illegal and anticompetive, yes. This is not a "they are big" problem, it's a "they're big and doing illegal price fixing with it".
Also, note that Google was caught intentionally deleting evidence they were ordered by the court to retain.
>I mean, the fact they abused that was illegal and anticompetive, yes. This is not a "they are big" problem, it's a "they're big and doing illegal price fixing with it".
(Note the separate case which determined Google is running an illegally anticompetitive operation in Search was a separate case which can be referred to as "United States v. Google LLC (2020)" and there is a third case they lost recently, Epic Games Inc. v. Google LLC, which determined Google operates an illegal monopoly with Android as well.)
I live in an urban area. Most people I know have a cleaner. Most of those people, including myself, found their cleaner via word of mouth. No services, no googling, no ads.
I mean, the person is looking for a cleaner in their area. If all of the cleaning businesses in the area slash their marketing budget to 0, the author is not going to fail to find a cleaner. All the marketing budget is doing is funneling people who want cleaning to one cleaning company over another.
> All the marketing budget is doing is funneling people who want cleaning to one cleaning company over another.
Yes, and anecdotally I've heard of better experiences using services that do not appear on the top search results. The reason being that the top results have already captured the local market and so are less incentivized to respond quickly, accept the job or task, or offer a better rate. They already have their business and may not need yours.
>If all of the cleaning businesses in the area slash their marketing budget to 0, the author is not going to fail to find a cleaner.
You're right, they'll find whatever incumbent cleaner instead. A marketing ban is something that all incumbents would love, because they don't need to attract more customers whereas marketing is basically the only way that upstarts can get a foothold.
When Google has the monopoly on marketing of cleaning companies in your area, from a consumer standpoint it’s effectively the same as if one cleaning company has the local monopoly. The way to win is to pay Google a bigger cut than your competitors, so Google just takes the incumbency premium as its marketing fee instead of the cleaning company.
If a buyer has access to the stored knowledge of trusted peers--peers who have knowledge of trustworthy sellers--supply can meet demand without involving an arms race between predatory middlemen.
The modern web was designed by predatory middlemen who want a cut of transactions they otherwise have no business being involved in. It's a textbook case of rentier capitalism.
So what needs to change is that we need to identify the design decisions made by those middlemen, rip them out root and branch, and fix the gaps with something that takes as an input the trust graph of the users so that the only way the middlemen can stay relevant is to personally gain the trust of each user whose transaction they've gotten in the middle of, and we need to publish the result as a protocol, not a platform, so it can be used without us (the authors of the protocol) being at risk of becoming the problem we're trying to solve.
I don't get it, is google blocking people from making or requesting word of mouth referrals? Or are people switching to google ads because it's more convenient? It just sounds like you're using "rentier capitalism" to describe companies you don't like.
Well yes, that is my main reason for disliking companies. And yes, google weighs in on browser standards in myriad ways which gives themseves and companies like them the ability to elevate the preferences of their customers above the preferences of their users.
It would be nice to block them from doing so, but the real fix is to give those users something better to use. Not much has gone into using technology to amplify the innate peer-to-peer trust/distrust mechanisms that we've spent millenia evolving such that they scale to the demands of our times, and quite a lot (thanks to google and friends) has gone into suppressing them.
>Well yes, that is my main reason for disliking companies. And yes, google weighs in on browser standards in myriad ways which gives themseves and companies like them the ability to elevate the preferences of their customers above the preferences of their users.
What does google's control over web standards have to do with the death of word of mouth referrals? You might not like FLoC or webusb but those aren't the reasons why everyone doesn't bother with word of mouth referrals to hire cleaners.
I'm watching a video sent to me by somebody I trust, and it stops to show me a video about the same topic made by somebody I don't trust, an interference which was targeted by--and an interference that I'm discouraged from preventing by--those standards. The connection is quite direct.
Now I don't know if there are any home cleaners that attempt to reach a wider audience on YouTube. Maybe there's a different medium that might suit their business better. But whatever it is, if it tries to be faster than meatspace gossip, there's some advertising platform selling the ability to interfere with it.
Google ads is a local optima for companies but not for consumers. The trouble is, for Google, the customers are the companies buying ads, not the people browsing the web. It's a classic example of not paying for your externalities
That Google isn't blocking a better model doesn't mean they aren't at fault. Ads are like pollution for our minds, we need a better web
>Google ads is a local optima for companies but not for consumers.
Are you sure you don't have it reversed? Companies would be quite happy if they could enter into some sort of no advertising pact so they don't have to spend any money on ads at all.
>The trouble is, for Google, the customers are the companies buying ads, not the people browsing the web. It's a classic example of not paying for your externalities
No, it's fully internalized, because consumers are getting free content (ie. sites where the ads are placed) and services (eg. gmail) in exchange. I'd be far more sympathetic to your claims of "externalities" if google stuffed its ads into your computer like junk mail makes its way into your mailbox.
> Are you sure you don't have it reversed? Companies would be quite happy if they could enter into some sort of no advertising pact so they don't have to spend any money on ads at all.
That's why it's a local optimum. Any company that try to unilaterally leave advertising will be punished. The global optimum would be no advertising at all, of course.
Anyway the people are already fighting back. I block ads everywhere, at least.
This is cool. I'm interested in reading more on this concept. Any chance you have a write up of experiences so far? Or can you point to other resources?
This looks fun, thanks for sharing. Will definitely give it a shot soon.
I read over the repo docs and was amazed at how clean and thorough it all looks. Can you share your development story for this project? How long did it take you to get here? How much did you lean on AI agents to write this?
Also, any plans for monetization? Are you taking donations? :)
I might write a short post on the development process, but in short:
- started development during Easter so roughly a month so far
- developed mostly using Cline and Claude 3.7
- inspired and borrowed heavily by Cline, Playwright MCP, and Playwright CRX which had solved a lot of the heavy lifting already - in a sense this project is those 3 glued together
I don't plan to monetize it directly, but I've thought about an opt-in model for contributing useful memories to a central repository that other users might benefit from. My main aim with it is to promote open source AI tools.
I've never used Claude Code or other CLI-based agents. I use Cursor a lot to pair program, letting the AI do the majority of the work but actively guiding.
How do you keep tabs on multiple agents doing multiple things in a codebase? Is the end deliverable there a bunch of MRs to review later? Or is it a more YOLO approach of trusting the agents to write the code and deploy with no human in the loop?
Multiple terminal sessions. Well written prompts and CLAUDE.md files.
I like to start by describing the problem and having it do research into what it should do, writing to a markdown file, then get it to implement the changes. You can keep tabs on a few different tasks at a time and you don't need to approve Yolo mode for writes, to keep the cost down and the model going wild.
> Customer support lines/banks/etc must shift their opening hours slightly so that they are open outside of business hours, so that working people can use their services
How would customer line/bank/etc workers ever get help with these services?
How do they get help with those services currently? At present they operate during the same hours those people are working in, shifting the hours makes no difference to them. So this plan changes nothing for them, but is beneficial for everyone else. I welcome suggestions on how to improve it for cusomer support reps too.
I've worked in a "first line of customer support" role, while simultaneously being a customer. Whenever I needed help with something, with my end-user hat on, I knew who to ask internally to get what I needed.
