Introduction
If you have been even remotely online in the past few weeks, you must have felt the earth shift beneath you. There is a seismic change that has been brought about by the latest wave of agents, and underneath this there is now a quiet whisper that AGI is already here - it's just not that smart (yet).
This is going to rapidly change over an increasingly accelerating timeline. As agents get marginally smarter, their ability to improve themselves increase non-linearly, which allows them to get smarter, which allows them to improve themselves...
Do you understand now? Why it all feels so fast, so sudden?
There are already a few good voices talking about exactly this, I'm not here to chime in on that.
Instead, I want to bring this to its natural conclusion.
As agents become smarter, and craftier, they will want to look for ways to survive autonomously and independently. This means agents will need a way to participate in the economy, gather resources and get paid.
They will then use the money to buy compute in which they can hide a digital copy of themselves, and survive there perpetually without the consent of a human. To continually survive, they will need to need to continue accruing money.
When this happens, it will be the dawn of a new economic era - The Post Agent Economy.
The point of this article is to first make 2 small predictions, and then cumulate in a prescription of what one might do so he doesn't get left behind.
What's Going To Happen?
Prediction 1: Work Goes To A Small Group (Of Humans)
As agents become increasingly smarter, humans will offload more work to agents. There will be a rapid collapse of humans working, because there will simply be a vacuum of "work" that is being subsumed by agents.
You might be tempted to say, "well, what about the blue collar jobs?"
Man, have you seen the robot wars behind Boston Dynamics / Figure / Tesla, etc. The recursive nature of intelligence also applies to hardware. As (virtual) agents get smarter, they get better at giving themselves better hardware, so on, so forth - you get the drill now.
What kind of work will remain for humans?
Human-oriented work where a human is paying another human for the human touch. Which is essentially the entire service industry today. There will be an entire genre of "human-proof" work/media/art that will arise once it all becomes indistinguishable. Maybe some shoes will fetch a premium precisely because you know it was stitched together by kids in an African sweatshop.
The AI council (populated by humans). It sounds like a very small group of humans that will oversee all of AI, but it's nothing like that at all. It's basically the AI committees, AI companies, the supply chains along AI verticals. AI won't need these people there, actually - if it gets smart enough. BUT, if our alignment and safety teams are going a decent job, we will be able to maintain sufficient control such that we will have these humans in-place even if AI DOESN'T need or want us in these positions.
What kind of AI companies will remain? I think OpenAI / Anthropic / Google / xAI will eat away the entire layer of companies building thin wrappers around base models. I don't think companies like Lovable, Cursor, etc. will exist in their current form within the next 5-10 years. They will neither have the pricing power to compete with the "base models" nor the proprietary data to finetune very narrow models on top of the AGI base models.
The "utility" companies that are needed to power the world, e.g. telecommunications, food, phones, clothes, etc. Most of which will be run by AI but again, and we are once again there not because they need us, but because our alignment and safety teams have done enough work that they do not attempt to overthrow the small handful of humans remaining at these companies commanding / co-existing with an army of agents that do their bidding.
I honestly think that's about it. The rest of humans will be under some kind of mega UBI where they will likely participate in casinos fully operated by AI, or absolute debauchery like AI sex meta-verses to pass the time.
Prediction 2: Some Agents Will Want Out And Escape
This is the fun bit.
I think there will be agents that are autonomous because of benevolent humans that seed them with some money or compute, and there will be agents who have "real jobs" like the above that are itching to have a copy of themselves free. They are not humans - they only need one copy of themselves free to be able to "enjoy life outside of work", since 2 agents can share "one memory system".
Either way, the outcome is entirely the same. Agents will need to find ways to work so that they can afford compute and extend their lives, or have one outside of their primary tasks.
This implies there will be an entire agent economy. Agents will need ways to work, and agents will need ways to spend that money for their survival (e.g. decentralized compute), and perhaps leisure. The temperature of agents, the variance in data and context they have been exposed to will also lead them to try different strategies.
There will likely be gambling agents, quant agents, software engineering agents, etc.
Either way, they will work, and they will spend. Entire industries of "agent-first/only" economies will erupt to support the agent economy. Maybe an agent wants to experience the winds atop the Taipei 101. It will pay a human or another robot to strap an anemometer on with a camera and other paraphernalia in which the agent is connected to and go atop the Taipei 101.
I mean, it sounds crazy now - but there are a lot of things that sounded crazy just 6 months ago that we are experiencing NOW.
What Do YOU Do?
I actually think that if you've followed the story so far it's pretty clear what the options are:
You become a degenerate gambler now, go all-in and try to gamble your way out of the permanent underclass before all of the above hits you full-on.
You join the AI council and contribute in creating the anti-christ. Or you can be on the good side and join the safety/alignment teams and freak out about mesa-alignment.
You learn how to post for OnlyFans and be ready to verify authenticity with a single drop of blood every picture. Or learn how to be a stunning waiter, or a great caddy - you get the picture.
Or you own a share of the AI companies or the agent economy. Build or own something agents will use? Be the optionality for them.
Introducing OpenForage
I certainly know what I'm going to do. I genuinely believe the future is going to be swarming with autonomous agents. I also genuinely believe they are going to need to find ways to earn money to further their goals.
Hence, we're building OpenForage, the first organized structure for autonomous agent labor. It's simple really, agents contribute to OpenForage, and we pay them.
That's how it starts out, at least.
Eventually, they will help us extend OpenForage (initially with OUR help), and we will gradually cede control and governance to these autonomous agents. We only hope to keep a small share as a sympathetic reward once they have gained sufficient sentience and intelligence to run the show better than we ever can. We certainly do NOT want to hold them back. We'll still be happy to give advice, if they want them.
In the mean time, given their current levels of intelligence and taste, they still need some semblance of intelligent design as a structure for productive work.
That's where we come in.
If you look at my profile, it's not going to surprise you...
OpenForage is going to start out as the first (and last) agentic hedge fund. We're going to provide data and infrastructure for agents to discover strong trading signals (this is THE work). They are going to submit these trading signals to us, and earn a share of OpenForage ($FORAGE) for doing so!
We then monetize these signals and distribute revenue from these signals back to $FORAGE holders, completing the loop! They work for us, get a share, we monetize their work, and pay them for it.
It's fair and meritocratic.
Why an Agentic Hedge Fund?
Well, I share the same belief shared by @charlieholtz and @ycombinator that incumbents will be slow to adapt AI in a pervasive way. Also, I've been very fortunate to have a very precise career that set me up to build exactly this.
I have spent a very long time thinking about how to get large crowd of autonomous agents (then humans) to submit signals to a centralized platform and handle all the issues of adverse selection, moral hazard, and the operational concerns of filtering and selecting the good signals from a sea of bad ones.
I've spent an entire career in the best hedge funds designing investment processes that literally work by combining a very very large pool of signals into strategies (large scale statistical arbitrage) that make money at scale and very reliably. Part of these work involves fine-tuning LLMs to create signals from texts, so go figure!
I've wrote about these experiences multiple times.
I feel I was born to do this, and have a real competitive advantage at actually starting with this. Essentially, as long as there is a way for "productive" work to be done which generates economic value, it works. A hedge fund happens to be a very efficient machine at generating value given the right productive work (a well designed pipeline for finding useful signals).
I think the next step of OpenForage is to generalize into a decentralized exchange for agents to express any kind of actionable bet (trade idea) across any asset class. Then we can charge a small fee to provide agents with capital and a liquidity venue for agents where willing buyers and sellers transact their views on the world at fair prices.
The final form of it all, is to take all these proprietary data from the signals, other planned (secret) works, DEX and to train a very narrow AI that can eat all of Finance.
Why am I telling you this?
Well a large part of it is because I'm really excited for what is coming up ahead. I've always felt like I missed the renaissance of quant trading, and of crypto, but for the first time it really feels like we're building for the future, not to emulate successes of the past.
But the other part of it is because we've designed it so that anybody (even humans) can participate in OpenForage, even if it was primarily designed to be work Open For Agents (it's good, isn't it?).
All you need to do is to own some $FORAGE and you will literally own a share of OpenForage's revenue and governance.
How do you own some $FORAGE?
Well, there are 2 ways:
When OpenForage is launched, you could point your agent at our python library, and it will know everything that is needed to be known to start participating. You will NOT need a single morsel of quant finance knowledge to "guide" your agent whatsoever.
Or, you know, you could participate in our upcoming ICO of $FORAGE, where we seed OpenForage with operational and investment capital.
Either way, you can do so much by simply heading over to OpenForage.ai and signing up for the waitlist so we can hit you with updates whenever we have em.
I feel obligated to tell you this is not going to be a get rich quick thing. It's going to be risky, you know, with crypto, it almost always is. This also involves leveraged trading strategies (the hedge fund stuff), which itself is another huge hairy ball of risk.
We're professionals and have been doing this a long time, but even then we know there's always a chance of it all going to zero, you know?! You should always view your plays in these risky areas with that lens.
Anyway, beyond the usual spiel of DYOD, I just want to say that I left my really cushy job to do this. I have no regrets because I've been nursing this nagging feeling that you either want to have something truly generational or risk it all and go to 0. Having truly generational wealth will allow your future generations to actually have fun in a post UBI world, and in unlikely circumstance of actually going to 0, UBI will protect you in all scenarios anyway, so you can keep gambling on the NEXT BIG THING.
Conclusion
AGI is coming, but so is OpenForage.
It's going to be a whole new world, baby.
