A serious point about this is that it's an AI that's an organization. This matters.
People laughed at Gas Town, and at that "Fences, not Sandboxes" article yesterday.[1] They're both AIs set up as an organization, not an emulated human. They're both expensive to operate, because the AI members of the organization spend a lot of time talking with each other. They're both applied to frivolous problems.
Now we have the same concept, applied to a more useful problem. It's time to pay attention. Something important is happening here.
Everybody has been talking about "human-level AI", or "superhuman-level AI". This is a different kind of AI - corporate-level AI. It works and thinks like a company's management. It watches a wide range of data sources and responds to some of them with action, and some with internal discussion within the AIs. It can potentially pay attention to far more inputs than any human. That overcomes a basic problem of corporate organization.
A classic management failure is not paying enough attention to something.
This is usually not a problem for computer systems.
Humans are bandwidth-limited. Teams of AIs are far less bandwidth-limited.
They may soon be outperforming humans for that reason alone.
Think about this for a while. These team AI systems may be very useful.
We have some interesting evidence from AI agents. People used to make the plan with a stronger and more expensive model, and then use a cheaper model to do the "grunt work" of implementing it.
But it turns out it's better to do it the other way around. A good plan is easy, implementing it is not.
Your point is well taken, but one thing stood out to me.
> A classic management failure is not paying enough attention to something. This is usually not a problem for computer systems.
This is actually my #1 problem with frontier LLMs. They are plenty smart, but they have no true long term memory, and their context window is not nearly enough for a complex system like a corporation. When my project gets big enough and I'm using a lot of /compact, the LLMs always have the same failure point: they fail to pay attention to something.
When they figure this out, I think it's going to be a very, very big moment.
> One of the miners, rejected on account of his size, offered to fight any man there as proof of his suitability as a soldier, and six men were eventually called upon to remove him.
Sometimes the claim is "set vision" - which AI can extract from the collective.
Someimes the claim is to prioritize the work (with bias), AI can extract that from the collective too, right?
Making resources available to the team? AI.
Coordinating between teams? AI.
Can the AI do this in a more un-biased way? Can it refine collective intelligence and desire into objective goals?
More than a software factory id love to see a manager-factory.
One thing, AI cannot do (yet) is bring the network. Like when you hire folk from Booth or Sloan so you can sell to places that hired Booth and Sloan grads.
Be friends with the board. Negotiate high pay and benefits (for themselves), and a golden parachute in case things go south (which they do most often). Take credit for things that were going to happen anyways. Blame xyz for things that didn't work out.
These are all impossible tasks for AI. AI can't play golf.
It’s going to be funny when the Tesla robots start showing up at tee times because the CEO couldn’t make it. Like a meeting agent taking notes for people who didn’t attend
> The job of an executive is: to define and enforce culture and values for their whole organization, and to ratify good decisions.
(And that's the whole list! Nowhere do you find set tasks or make decisions)
To me, that means that they're still pretty essential - on both fronts, AI needs values and AI needs a system that can tell whether decisions are good or not.
If the operation is to distill options into a few choices, and then pick the "best" one then, I think hivemind can win there too. Real-time democracy running a business.
For example an open-source project could/would still have a BDFL but, the AI-CEO could do the operational busywork while the BDFL is out talking to humans and presenting at PGCon or some other productive thing.
Nope! that's the subtlety. "distill options" is not the CEO's (or any executive's) job.
The way he phrases it, if I remember correctly (and this is by no means the only way to executive, I just really like his framing), is that the way you make decisions is by getting your immediate subordinates to agree, and they iterate. Decisions should be made at the lowest level that they can be made at, and if that's the CEO level, you've made mistakes.
So you make the values list, you make sure that people live by it (up to and including firing people who don't and hiring people who do, which he basically thinks is the whole day to day), and then you bring your subordinates together to agree on decisions. If they don't agree, you send them off to work through it together and come back with consensus (or, less great, compromise). If you have to put your thumb on the scale to get a good decision, again, you're not CEOing well.
Business actually treats a pseudo-dictatorship (with the requisite yes-men) as "good leadership".
Most times it's pretty simple, in my experience. A CEO either runs a business as a (1) Sales focus [best deals and vendors] or (2) Marketing focus [best reputation and trends] or (3) Investor focus [best boardroom friends].
Lots of other opinions are usually thrown out until there's really only one facet used for running the company.
At my current job, a 10 person startup, the CEOs job is to generate leads, talk to potential customers, and close deals. He, along with the other founders, shape what we work on based on what he thinks it’s possible for us to sell. He’s essential to the business.
At my previous 50k person company I never talked to anyone within 4 levels of the CEO and thought he was useless. But realistically he probably had a somewhat similar job - talking with the board, talking with politicians, and talking with the CEOs of other companies in the effort to lower costs (tax breaks, partnerships, etc) and generate sales (contracts, partnerships, etc).
Similar, yes, but instead of finding out who else can give you money at 10 people, they have to find out who else they have to give money at 10k people. Leads and lobbies.
Marketing and they hire and fire important people. That's about it though, where we should take the product, what features to do next, partnerships, these are pretty obvious to anyone paying attention.
I don't think these things are obvious. I've worked at many places that have had big, company-changing forks in the road available to them and navigating those forks wasn't easy (and in my opinion several times we chose wrong). The crowd wasn't always right or even well aligned and sometimes the only agreement was that a middle of the road approach would be worse. My intuition says getting that right takes a lot of luck, good momentum, and a lot of departments working well together (bringing in and communicating market info, broadcasting marketing that works, making good partnerships, etc). The cynic in me thinks one big impact of a ceo is the various cheating and corruption tools that can make some of those things easier or at least less critical to get right (you don't need to make good cybertrucks or convince people to buy them if the government agrees to buy 450k of them as "armored vehicles", for example).
If that kind of thing is easy for you (the honest ones or the clandestine ones) then you might make a really strong entrepreneur/business leader.
> Sometimes the claim is "set vision" - which AI can extract from the collective.
To be an effective CEO, you have to make judgement calls that the collective can’t always have wisdom and context on, is confidential, or requires fast decision making during which there are interactive discussions or negotiations!
This is highly subjective and a gives an impression that every one is doing a very effective job. The reality is much more grey when you account for so many companies that get shut down.
My personal take, somewhere down the line people feel scared of taking responsibility for more than themselves and their family. This limits the ability for most people to take on such ventures. Not everyone falls in the same bucket but a lot of leadership in todays date is about taking the role without responsibility. They happen to be in the role with the help of network affect and somehow that just amplifies.
> This is highly subjective and a gives an impression that every one is doing a very effective job. The reality is much more grey when you account for so many companies that get shut down.
I believe that the average are average.
I don’t believe what I wrote suggests that. I was attempting to be as neutral as possible.
CEO's exist as a sink for liability. When something goes wrong, they are the fall guys, that's ultimately their job. That's why CEO compensation is generally correlated strongly to earnings and they usually have golden parachutes.
Without a CEO to fire when the business goes south, how is the board supposed to respond? Taking accountability? lmao
I deliberately stopped short of invoking golden parachutes and the numerous other escape hatches available at that altitude, as I am perfectly aware they exist. I am equally aware of chief executives who have departed with remarkable agility from organisations they themselves had first steered off the cliff.
My point, however, remains the same: as the letter of the law presently stands, AI can't be made accountable – at least not today.
LLCs and other corporate structures only (mostly) protect employees and shareholders from civil judgments. CEOs and other executives are also at risk of criminal penalties for Sarbanes-Oxley (SOX) violations and similar. Actual criminal prosecutions of CEOs have been few, but not zero.
The answer is to be responsible to the business. The buck stops at the CEO. Therefore, their responsibilities are technically everything in the entire company. And I think many people who have only worked for other people underestimate how valuable that responsibility is.
Many people observe a limited subset of reality. It's highly visible when a company hits a crisis and people lose their jobs; it's much more common, but much less visible, for a CEO to get an after-hours call about a brewing problem which they must immediately work to smooth over.
I get it’s a joke, and often a frequent cynical joke made here, but I actually think it’s a good thought experiment. Almost like a new Turing test
The thought experiment, to be clear, is building an artificially intelligent CEO or company founder with the intent of generating profit. If this AI can turn a profit, it has passed this new test
Of course, by the time this new AI starts turning a profit, it will be maximized to the extent that it loses its competitive advantage, and then we’re back to square 1. That is, when everyone uses the same AI, nobody has a competitive advantage over anyone else
My current theory is that because of this, it is not possible to build an AI that generates a profit perpetually. Only temporarily. Then human intervention would be required to adjust things to make the AI competitive again. But at that point, it’s no longer an AI driving the profit, it’s its human tinkerer
Doesn't the incentive here, lead to privatizing the AI and possibly releasing a different product to the public - or at least with a delay? (not too unlike how the frontier labs work today).
Yes I would think that would be the endgame goal. But in practice, I’m very skeptical a purely AI-driven executor can stay competitive when the market and government supporting it and surrounding it is driven by humans
You’re presuming here that CEOs cause companies to make profit instead of serving as a scapegoat for their often unavoidable market dynamics (at best).
At what point does a company become a company if one person doesn’t make a company? What’s the threshold for non-labor where a company becomes a company?
“Walking away costs you the deal. Signing costs you the company. The math is obvious in retrospect; harder in the moment when there's quota pressure”
such valuable, should I say load-bearing insights
Other gem from my new CFO:
“The financial statements are the result. The variance analysis is where the operating insight lives.”
And my new C(3-)PO just told me
“If we removed all sales involvement, what % of revenue would we still get?
#5 is the honest PLG test. If the answer is <30%, PLG is supplementary at best. If the answer is >60%, PLG is the actual motion and the rest is overlay.”
Searching for PLG, I found: “In short, companies that use PLG models gain another revenue-generating channel: the company’s product itself.” What a world we live in, that a company's products being a revenue-generating channel is an innovation worthy of note!
The only thing worse than a human CEO is an AI CEO with power to fire you. People just don't think it will do stupid things, have nonsensicle orders which they have to follow, and likely fire people left right and center.
But oh ho there is no human ceo, must be a good place to work!!
Servant leadership is a management cliche that works well as a counterbalance to the dominance hierarchy that's implicit in a human org chart. I'm not going to listen to a Claude instance telling me to do something I'd prefer not to do.
They’re just doing what they’re told to do by the shareholders. They’re the largest cogs sure, but cogs nonetheless, and they’re replaced just as quickly when they fail to make choices that maximize profit above all else.
This reminds me of William Gibson's "Burning Chrome", where the head of a syndicate is a Russian AI (Chrome) running things behind a smuggling front. Gibson basically predicted the AI CEO forty years early!
If human-led companies started consistently losing to AI led companies... would it say more about the intelligence of AI, or about how thoroughly power corrupts humans?
I think it's pretty fun. Worker cooperatives no longer need different ranks of executive officers. They can all be peers with different functions. So if worker cooperatives are going to succeed in the US, they have pretty good conditions right now.
I've been slowly developing my hermes agent to do something similar but is very customized towards my needs. I should spin up another agent and give this one a shot and see what it's capable of. This is hilarious.
To the extent this is tongue in cheek it is funny, but I could imagine if I was building a startup I’d like something like this if it was good.
That said, if you look at what it’s actually doing (read the question carefully, then read what it actually said) I am not convinced this is better than what ChatGPT could do out of the box already.
A team of software developers were recently let go so a company could replace them with AI. So they got together and created an Open Source AI CEO to replace the CEO and other executives. Hopefully, turnabout is fair play and this might even get some folks to think twice about using AI to replace people.
Corporate officers, including CEOs, have to be "natural persons" (legal jargon for flesh-and-blood humans) either directly by law or indirectly as a practical matter so the attempted turnabout falls flat on its face.
Sun Tzu said to "know your enemy" and those who don't know the basics of how businesses and finance work have no hope of defeating them.
I’d be more interested to see a collection of independent guild, unions, and co-ops that voluntarily coordinate through an AI intermediary.
Because the AI can’t really own things or be held responsible, the idea of a skeuomorphic AI seems quite silly. Keep the ownership in the humans, but plural, and have them in contact with the actual work.
This is one of those rejoinders that tries to be clever but really, really isn't. The "intern as fake CEO" is left holding the bag because it's their signature on contracts and other legal documents. The idea of puppet CEOs to fraudulently dodge accountability existed well before LLMs became a thing. Courts see right through that kind of scheme plus the "intern" is likely to turn coat against the mastermind to save their own skin at the first sign of trouble.
If you think CEOs are the primary legal signatories for their company you might need to spend some time away from garage band startups. Companies maintain entire legal departments for a reason.
This is hilarious, I feel like we are going to see AI companies add something to the system prompt or terms of service that explicitly forbids having an LLM act as CEO. Honestly though I do hope this gives pause to the C suite when it comes to replacing people with AI
We've replaced our CEO for the better part of 6 weeks now. He has been on a media tour promoting a product we no longer sell, but it keeps him busy while we continue to grow.
While Proxy AI [0] is not open source, it is free, it is built from pure garage energy, and we are having our summer sale right now. 20% off for HN users.
I think AI CEOs are dumb but AI 'enterprise decision-makers' could be very very interesting.
Imagine how quickly your meetings could get resolved if you had teams submit prompts / contexts, had a clear, documented set of overarching objectives, etc.?
We're all joking around here. But 5 decades down the line, with increasing cognitive offload to AI, is a slow erosion leading to this end state so crazy?
Many CEOs are bad and their businesses will either fail or be bailed out. But the work CEOs do is probably some of the last to get automated - higher order strategy, relationship building, managing people.
If you want to see that idea extrapolated to an impressive extent, check out Accelerando by Charles Stross. If Stross was right we're in for a bumpy ride.
Highly configurable. Looks like it is targeted at any level of management. The key features seems that it can generate its own vision of how to move something forward and that it can be set with degrees of freedom depending on the use case. Video demo here: https://www.youtube.com/watch?v=O_g97xxVTMk
Everyone seems to be missing the point of the role of the CEO.
The CEO is not some masterful person answering all the questions for the company, I would argue that a CEO should be able to disappear for a year and if they had done their job correctly and assembled a good team the company would do just fine without them.
No the purpose of the CEO is to safeguard the Capital that has been invested in the company. That is all. They make sure it doesn't walk our the door in wages or other wasteful spending. They are the Capital's man on the ground.
Yes, this is actually a reason why an AI executive could be a formidable alternative in theory.
The CEO is ultimately an employee of the board, and managing the CEO with respect to the board's oversight and interest is a somewhat adversarial challenge.
However, the AI CEO's intentions are fully transparent to the board and entirely pliable to the board's demands. The AI CEO also offers the board the opportunity to retain significant equity shares that materially impact their wealth - in this way AI executives stand to be an even more tantalizing target than fixed labor costs.
Of course, this presumes LLMs will become reliable enough for this, which doesn't seem likely.
I would argue that many companies would in fact be better off disappearing for a year, rather than constantly thrashing the company to chase the latest trend.
And judging from the ZIRP hiring sprees and tokenmaxxing, CEOs are more often the driving force behind wasteful spending than they are the safeguards against it...
Op here. Team was let go. Decided to do a startup. We thought that tools like co-pilot and Claude code were transforming development. We wanted to transform management (not just CEOs). We open sourced this knowing that if people find this useful (and we hope they do) we can get a support work out of it (similar to the Redhat model and other open source tech).
I’m all for productive people, brilliant engineers. I still get dirty with codes/work all the way down to the newly hired fresh out of college kids, to the production architects that shapes multi-million, if not billions, of P&L. I’m also very lucky to be included in key decision making, planning, negating amongst the boards of C-Suites, Directors, VPs.
A very personal view but many developers/engineers fail to see the overall picture. And yes, I was there and I went through all of the mistakes.
The things you see on the surface of those incompetent CxOs are just the tiny piece of the bigger picture. You cannot replace blames, decision in the face of uncertainty. Almost all decisions that businesses take have nothing to do with the proficiencies of technologies or engineering finesse, or even the best options, but mostly on “incentives” and who can take/own the “blame.”
“AI” replacing developers is evident because that is a rather tiny perspective and kinda easy to do the equation - developer can do that (good, bad, best) and can the AI do it - maybe, perhaps. So, replace.
A C-Suite took a decision to tick the final approval of a billion dollar deal with a service providers (amongst 10s of them). The idea might have sounded idiotic because there were 3 others better at the top than the chosen one. Even if all of the task and options were weighted with AI, but that last human decision is the Blame that the C-Suite took on, by betting his incentives on it (money, fame, pride, ego.)
> Almost all decisions that businesses take have nothing to do with the proficiencies of technologies or engineering finesse, or even the best options, but mostly on “incentives” and who can take/own the “blame.”
This argument always confuses me because it seems like lower level employees are the ones that get quickly dismissed when things get bad while a lot of C level executives seem to have some sort of compensation even if they fuck up.
I think a lot of this kind of talk treats the C-Suite as special, mystical "visionary" people who Know Things We Can't Know, and by virtue of this otherworldly magical power, are uniquely qualified to do their jobs. Like the average person with basic knowledge of his business and industry can't possibly choose one service provider out of 10 bids, or set business priorities, or drive and enforce culture. We're not talking about wizardry here.
People always dunk on me for saying this outrageous statement, but if you put me in charge of a random Fortune 500 company, I give myself a 50% probability of the company performing better than if its real CEO was in charge, and a 50% probability of it performing worse. I really don't think these guys bring as much to the table as they think they do and as much as the general public seems to think they do. Business success does not come from paranormal insight and abilities.
I think you could randomly nominate CEOs from the set of the general population with, say, a minimum of a college degree, and the randomly nominated CEO's performance would not statistically differ from "qualified" CEOs.
This is absolutely spot on. CEO/C-Suite/VPs are surrounded by gatekeepers (usually something like the PMO org, program management) who feed them with garbage info. Even if a top level leader is genuinely interested in making things better, they would never know because the gatekeepers firewall any competent person.
The gatekeeping org has the highest density of incompetence. They only hire incompetent because loyalty to the person hired is the most important. Gatekeeping orgs of various leaders protect each other, there is almost never any blame, it is things are good and they are going to get better, announce xyz initiatives/projects.
> A very personal view but many developers/engineers fail to see the overall picture. And yes, I was there and I went through all of the mistakes.
What is failling to see the overall picture? I hate this overall picture mysticism. I bet that if people where given the same role and information as the CEO they would be able to see the overall picture.
The difference is that the CEO controls the capital and has the ear of the investors, whereas the developers generally don't. It's the investors that control whether the CEO is a bot or not.
Honestly the best use of AI at the moment is to replace non-technical middle management layers. Many of those yelling the loudest for “more AI now” are quite ironically holding the jobs best replaced by the current creation of AI.
Number of people who think that a leader or a CEO can be replaced by AI and who themselves have actually risked everything to start a company, be a leader, or do anything similar: exactly 0.
> Number of people who think that a leader or a CEO can be replaced by AI and who themselves have actually risked everything to start a company, be a leader, or do anything similar: exactly 0.
Prove it.
> be a leader
There are a lot of bad "leaders" out there running Fortune 500s, fwiw.
If you just want a sociopath with faux accountability that's not very difficult, dime a dozen.
This has likely been the case as long as jobs have existed. Some CEOs spend their entire time tweeting, apparently.
The way any large company works, and this includes all the big tech companies, is that they end up in a state of constant churn. Every 6 months or so there's a reorganization. There'san old joke that you get an email that your manager's manager's manager's manager now reports to a new manager and you've never heard of any of these people.
The point of that churn is to make sure nobody is ever held responsible for something not working. It's never given enough time. That's by design. Middle (and even senior) management is about empire-building.
Another quote that springs to mind is that the bureaucracy is expanding to meet the needs of the expanding bureaucracy. That's how I feel about management bloat.
Anyone who has paid attention to Intel's downfall in the last decade will have heard stories about how the company descended into fiefdoms and it became impossible to get anything done. That's what happens.
Most people in these organizations aren't particularly talented at anything other than forming social relationships. the leadership of these companies always end up having an adversarial relationship with their workers and try to suppress their wages to increase profits so they can get their stock options and bonuses and the CEOs all sit on each other's boards of directors setting each other's compensation packages. It's a lootocracy.
The point of all this is that the leadership will never decide that they are the problem. Take Meta's Metaverse. It was a $70 billion flop. A lot of people got laid off from that. Were any of those worker bees responsible? Nope. They didn't choose the strategy or probably even what they worked on. It takes a lot for heads to roll within senior leadership and with the vote-stacked share structures of many major tech companies, ousting a CEO is essentially impossible.
A serious point about this is that it's an AI that's an organization. This matters.
People laughed at Gas Town, and at that "Fences, not Sandboxes" article yesterday.[1] They're both AIs set up as an organization, not an emulated human. They're both expensive to operate, because the AI members of the organization spend a lot of time talking with each other. They're both applied to frivolous problems.
Now we have the same concept, applied to a more useful problem. It's time to pay attention. Something important is happening here.
Everybody has been talking about "human-level AI", or "superhuman-level AI". This is a different kind of AI - corporate-level AI. It works and thinks like a company's management. It watches a wide range of data sources and responds to some of them with action, and some with internal discussion within the AIs. It can potentially pay attention to far more inputs than any human. That overcomes a basic problem of corporate organization.
A classic management failure is not paying enough attention to something. This is usually not a problem for computer systems. Humans are bandwidth-limited. Teams of AIs are far less bandwidth-limited. They may soon be outperforming humans for that reason alone.
Think about this for a while. These team AI systems may be very useful.
[1] https://news.ycombinator.com/item?id=49423146
It's funny that when you offload the work to agents they still find away to monopolise each other's time with meetings.
> This is a different kind of AI - corporate-level AI.
So the responses will be even more incoherent than Opus 5?
We have some interesting evidence from AI agents. People used to make the plan with a stronger and more expensive model, and then use a cheaper model to do the "grunt work" of implementing it.
But it turns out it's better to do it the other way around. A good plan is easy, implementing it is not.
Your point is well taken, but one thing stood out to me.
> A classic management failure is not paying enough attention to something. This is usually not a problem for computer systems.
This is actually my #1 problem with frontier LLMs. They are plenty smart, but they have no true long term memory, and their context window is not nearly enough for a complex system like a corporation. When my project gets big enough and I'm using a lot of /compact, the LLMs always have the same failure point: they fail to pay attention to something.
When they figure this out, I think it's going to be a very, very big moment.
Based on my observations, the most important characteristics of a CEO is to be over 6ft, and to walk about looking important.
CEO jobs are safe until the humanoid robots have mastered this skill.
I'm under 6ft and have never been a CEO, so this appears to check out.
Also, it has 4 legs, so it must be a cat.
My SVP is 6'4". 6 measly feet aint cutting it, that's just "king of manlets" territory.
Reject ‘manlet’ — RETVRN to ‘bantam’ https://en.wikipedia.org/wiki/Bantam_(military)
> One of the miners, rejected on account of his size, offered to fight any man there as proof of his suitability as a soldier, and six men were eventually called upon to remove him.
You’d think culture and personality fit with a board of directors would be an easier feat for a robot.
I think either gray hair or bald does a lot too.
Shit, why ain't I a CEO, then?
Based on my observations, ahem research, you must not have fired developers to make room for AI. Therefore, you are not qualified to be a CEO.
Using HN instead of walking around looking important
What does a CEO(and other managers) do really?
Sometimes the claim is "set vision" - which AI can extract from the collective.
Someimes the claim is to prioritize the work (with bias), AI can extract that from the collective too, right?
Making resources available to the team? AI.
Coordinating between teams? AI.
Can the AI do this in a more un-biased way? Can it refine collective intelligence and desire into objective goals?
More than a software factory id love to see a manager-factory.
One thing, AI cannot do (yet) is bring the network. Like when you hire folk from Booth or Sloan so you can sell to places that hired Booth and Sloan grads.
> What does a CEO(and other managers) do really?
Be friends with the board. Negotiate high pay and benefits (for themselves), and a golden parachute in case things go south (which they do most often). Take credit for things that were going to happen anyways. Blame xyz for things that didn't work out.
These are all impossible tasks for AI. AI can't play golf.
Chinese humanoid robots are 3 years out from playing a decent round on a par 3 course.
It’s going to be funny when the Tesla robots start showing up at tee times because the CEO couldn’t make it. Like a meeting agent taking notes for people who didn’t attend
To paraphrase Andy Grove:
> The job of an executive is: to define and enforce culture and values for their whole organization, and to ratify good decisions.
(And that's the whole list! Nowhere do you find set tasks or make decisions)
To me, that means that they're still pretty essential - on both fronts, AI needs values and AI needs a system that can tell whether decisions are good or not.
> AI needs values and AI needs a system that can tell whether decisions are good or not
Well, so do most CEOs.
There are hundreds of thousands of CEOs of companies you've never heard of.
If the operation is to distill options into a few choices, and then pick the "best" one then, I think hivemind can win there too. Real-time democracy running a business.
For example an open-source project could/would still have a BDFL but, the AI-CEO could do the operational busywork while the BDFL is out talking to humans and presenting at PGCon or some other productive thing.
Nope! that's the subtlety. "distill options" is not the CEO's (or any executive's) job.
The way he phrases it, if I remember correctly (and this is by no means the only way to executive, I just really like his framing), is that the way you make decisions is by getting your immediate subordinates to agree, and they iterate. Decisions should be made at the lowest level that they can be made at, and if that's the CEO level, you've made mistakes.
So you make the values list, you make sure that people live by it (up to and including firing people who don't and hiring people who do, which he basically thinks is the whole day to day), and then you bring your subordinates together to agree on decisions. If they don't agree, you send them off to work through it together and come back with consensus (or, less great, compromise). If you have to put your thumb on the scale to get a good decision, again, you're not CEOing well.
> Real-time democracy running a business
This is not a competitive advantage
Neither is dictatorship?
Business actually treats a pseudo-dictatorship (with the requisite yes-men) as "good leadership".
Most times it's pretty simple, in my experience. A CEO either runs a business as a (1) Sales focus [best deals and vendors] or (2) Marketing focus [best reputation and trends] or (3) Investor focus [best boardroom friends].
Lots of other opinions are usually thrown out until there's really only one facet used for running the company.
It is.
The best products are usually from single minds with a strong vision. "Too many cooks in the kitchen" is a phrase for a reason.
Your least favorite products are designed by committee or a vision less person following trends.
>AI needs a system that can tell whether decisions are good or not.
So they need someone to review AI output huh?
At my current job, a 10 person startup, the CEOs job is to generate leads, talk to potential customers, and close deals. He, along with the other founders, shape what we work on based on what he thinks it’s possible for us to sell. He’s essential to the business.
At my previous 50k person company I never talked to anyone within 4 levels of the CEO and thought he was useless. But realistically he probably had a somewhat similar job - talking with the board, talking with politicians, and talking with the CEOs of other companies in the effort to lower costs (tax breaks, partnerships, etc) and generate sales (contracts, partnerships, etc).
Similar, yes, but instead of finding out who else can give you money at 10 people, they have to find out who else they have to give money at 10k people. Leads and lobbies.
Marketing and they hire and fire important people. That's about it though, where we should take the product, what features to do next, partnerships, these are pretty obvious to anyone paying attention.
I don't think these things are obvious. I've worked at many places that have had big, company-changing forks in the road available to them and navigating those forks wasn't easy (and in my opinion several times we chose wrong). The crowd wasn't always right or even well aligned and sometimes the only agreement was that a middle of the road approach would be worse. My intuition says getting that right takes a lot of luck, good momentum, and a lot of departments working well together (bringing in and communicating market info, broadcasting marketing that works, making good partnerships, etc). The cynic in me thinks one big impact of a ceo is the various cheating and corruption tools that can make some of those things easier or at least less critical to get right (you don't need to make good cybertrucks or convince people to buy them if the government agrees to buy 450k of them as "armored vehicles", for example).
If that kind of thing is easy for you (the honest ones or the clandestine ones) then you might make a really strong entrepreneur/business leader.
Managers coach and develop people. Helping them grow, nurturing their strengths and connecting them with opportunities that help them achieve more.
At least the good ones do.
> Sometimes the claim is "set vision" - which AI can extract from the collective.
To be an effective CEO, you have to make judgement calls that the collective can’t always have wisdom and context on, is confidential, or requires fast decision making during which there are interactive discussions or negotiations!
This is highly subjective and a gives an impression that every one is doing a very effective job. The reality is much more grey when you account for so many companies that get shut down.
My personal take, somewhere down the line people feel scared of taking responsibility for more than themselves and their family. This limits the ability for most people to take on such ventures. Not everyone falls in the same bucket but a lot of leadership in todays date is about taking the role without responsibility. They happen to be in the role with the help of network affect and somehow that just amplifies.
> This is highly subjective and a gives an impression that every one is doing a very effective job. The reality is much more grey when you account for so many companies that get shut down.
I believe that the average are average.
I don’t believe what I wrote suggests that. I was attempting to be as neutral as possible.
Like RTO mandates which were definitely made with wisdom.
Not all CEOs did that. Just like all software engineers aren't great or excellent, all CEOs aren't. But to categorically denounce seems rigid.
They manage people. They’re people people.
CEO's exist as a sink for liability. When something goes wrong, they are the fall guys, that's ultimately their job. That's why CEO compensation is generally correlated strongly to earnings and they usually have golden parachutes.
Without a CEO to fire when the business goes south, how is the board supposed to respond? Taking accountability? lmao
Well, as soon as they start being replaced is there a network to bring.
In the current legal framework, AI can't be held accountable, a human being is required.
That is what the CEO is for – to be on the receiving end of beatings and flogging if the business they are at the masthead of effs something up.
Good point. CEOs of companies are often held personally accountable for corporate malfeasance by "the current legal framework".
I understand your sarcasm but that’s what the limited in LLC is for.
Unbounded liability means nobody starts a business, or they do it under the table without responsibility or taxes.
Turns out we want successful private businesses much more than we want to hold a few nasty CEOs accountable.
I agree. I just thought it was funny that someone was pointing out the legal unaccountability of an AI in this context.
There is no need for two penny sarcasm.
I deliberately stopped short of invoking golden parachutes and the numerous other escape hatches available at that altitude, as I am perfectly aware they exist. I am equally aware of chief executives who have departed with remarkable agility from organisations they themselves had first steered off the cliff.
My point, however, remains the same: as the letter of the law presently stands, AI can't be made accountable – at least not today.
The sarcasm was pretty funny so it has some utility, no need to be so serious
LLCs and other corporate structures only (mostly) protect employees and shareholders from civil judgments. CEOs and other executives are also at risk of criminal penalties for Sarbanes-Oxley (SOX) violations and similar. Actual criminal prosecutions of CEOs have been few, but not zero.
https://archives.fbi.gov/archives/oklahomacity/press-release...
> What does a CEO(and other managers) do really?
The answer is to be responsible to the business. The buck stops at the CEO. Therefore, their responsibilities are technically everything in the entire company. And I think many people who have only worked for other people underestimate how valuable that responsibility is.
Many people observe reality, where the workers end up being the ones held accountable for the CEO's decisions, not the CEO.
Many people observe a limited subset of reality. It's highly visible when a company hits a crisis and people lose their jobs; it's much more common, but much less visible, for a CEO to get an after-hours call about a brewing problem which they must immediately work to smooth over.
Workers are not legally or financially accountable for any of that. That is the whole point of a "job". I think you are confused.
When the worker loses their jobs over bad executive decisions, they sure feel like they were pretty harshly financially penalized for those decisions.
Neither are CEOs in our present society
Are you confusing CEOs with founders, shareholders, etc.? Chief Executive Officer is, in fact, a "job."
I wish that was true some how. In real world unfortunately the buck stops at whoever is easy scape goat, at times it can be a CEO
Rewponsibility is why we pay external consultants!
I get it’s a joke, and often a frequent cynical joke made here, but I actually think it’s a good thought experiment. Almost like a new Turing test
The thought experiment, to be clear, is building an artificially intelligent CEO or company founder with the intent of generating profit. If this AI can turn a profit, it has passed this new test
Of course, by the time this new AI starts turning a profit, it will be maximized to the extent that it loses its competitive advantage, and then we’re back to square 1. That is, when everyone uses the same AI, nobody has a competitive advantage over anyone else
My current theory is that because of this, it is not possible to build an AI that generates a profit perpetually. Only temporarily. Then human intervention would be required to adjust things to make the AI competitive again. But at that point, it’s no longer an AI driving the profit, it’s its human tinkerer
Doesn't the incentive here, lead to privatizing the AI and possibly releasing a different product to the public - or at least with a delay? (not too unlike how the frontier labs work today).
Yes I would think that would be the endgame goal. But in practice, I’m very skeptical a purely AI-driven executor can stay competitive when the market and government supporting it and surrounding it is driven by humans
You’re presuming here that CEOs cause companies to make profit instead of serving as a scapegoat for their often unavoidable market dynamics (at best).
In a 1-person company, who or what generates a profit?
One person isn't a company. Legally maybe, but there's no point in a company if you can profitably do all the labor yourself.
At what point does a company become a company if one person doesn’t make a company? What’s the threshold for non-labor where a company becomes a company?
I'd start with division of labor as a minimum.
You're describing the Tendency of the Rate of Profit to Fall[1], which, I'm sorry, is a Marxist idea so it's automatically wrong.
1. https://en.wikipedia.org/wiki/Tendency_of_the_rate_of_profit...
CEO here, thanks just fired all my C suite team,
https://github.com/SenteLabsAI/OpenExecutive/blob/main/packa...
“Walking away costs you the deal. Signing costs you the company. The math is obvious in retrospect; harder in the moment when there's quota pressure”
such valuable, should I say load-bearing insights
Other gem from my new CFO:
“The financial statements are the result. The variance analysis is where the operating insight lives.”
And my new C(3-)PO just told me
“If we removed all sales involvement, what % of revenue would we still get? #5 is the honest PLG test. If the answer is <30%, PLG is supplementary at best. If the answer is >60%, PLG is the actual motion and the rest is overlay.”
Searching for PLG, I found: “In short, companies that use PLG models gain another revenue-generating channel: the company’s product itself.” What a world we live in, that a company's products being a revenue-generating channel is an innovation worthy of note!
The only thing worse than a human CEO is an AI CEO with power to fire you. People just don't think it will do stupid things, have nonsensicle orders which they have to follow, and likely fire people left right and center.
But oh ho there is no human ceo, must be a good place to work!!
Awesome. There's no reason why AI can't replace all these highly paid CEOs along with the rest of us. Looking forward to the Delamain future.
Do you think that they will offer the Excelsior package for sale?
The best use of AI is to replace CEOs, C-Suite, politicians and bureaucrats. Let them feel the heat for the first time in their life.
Interested folks can start by putting in shareholder resolutions with the proposal and the prompts that would replace CEO. SKILLs.md?
So we work for the AI? AI should work for us.
Servant leadership
Servant leadership is a management cliche that works well as a counterbalance to the dominance hierarchy that's implicit in a human org chart. I'm not going to listen to a Claude instance telling me to do something I'd prefer not to do.
Feel the heat? As in, be punished? For what, exactly?
Wage theft
They’re just doing what they’re told to do by the shareholders. They’re the largest cogs sure, but cogs nonetheless, and they’re replaced just as quickly when they fail to make choices that maximize profit above all else.
For being egotistical psychopaths
No the best use is to replace hacker news commentors. Let them feel the heat for the fist time in their life.
See you can add anything there and the sentence will make sense.
I will never vote for an AI system to actually make laws lol wtf are you even saying
This reminds me of William Gibson's "Burning Chrome", where the head of a syndicate is a Russian AI (Chrome) running things behind a smuggling front. Gibson basically predicted the AI CEO forty years early!
A warning: Don't burn Chrome.
If human-led companies started consistently losing to AI led companies... would it say more about the intelligence of AI, or about how thoroughly power corrupts humans?
This is brilliant! Immediately forwarded to the boss.
I think it's pretty fun. Worker cooperatives no longer need different ranks of executive officers. They can all be peers with different functions. So if worker cooperatives are going to succeed in the US, they have pretty good conditions right now.
An auditable executive would certainly be a first.
I've been slowly developing my hermes agent to do something similar but is very customized towards my needs. I should spin up another agent and give this one a shot and see what it's capable of. This is hilarious.
To the extent this is tongue in cheek it is funny, but I could imagine if I was building a startup I’d like something like this if it was good.
That said, if you look at what it’s actually doing (read the question carefully, then read what it actually said) I am not convinced this is better than what ChatGPT could do out of the box already.
https://youtu.be/O_g97xxVTMk
A team of software developers were recently let go so a company could replace them with AI. So they got together and created an Open Source AI CEO to replace the CEO and other executives. Hopefully, turnabout is fair play and this might even get some folks to think twice about using AI to replace people.
Corporate officers, including CEOs, have to be "natural persons" (legal jargon for flesh-and-blood humans) either directly by law or indirectly as a practical matter so the attempted turnabout falls flat on its face.
Sun Tzu said to "know your enemy" and those who don't know the basics of how businesses and finance work have no hope of defeating them.
I’d be more interested to see a collection of independent guild, unions, and co-ops that voluntarily coordinate through an AI intermediary.
Because the AI can’t really own things or be held responsible, the idea of a skeuomorphic AI seems quite silly. Keep the ownership in the humans, but plural, and have them in contact with the actual work.
Neat. Hire an intern to prompt the "CEO" as needed. Problem solved.
This is one of those rejoinders that tries to be clever but really, really isn't. The "intern as fake CEO" is left holding the bag because it's their signature on contracts and other legal documents. The idea of puppet CEOs to fraudulently dodge accountability existed well before LLMs became a thing. Courts see right through that kind of scheme plus the "intern" is likely to turn coat against the mastermind to save their own skin at the first sign of trouble.
If you think CEOs are the primary legal signatories for their company you might need to spend some time away from garage band startups. Companies maintain entire legal departments for a reason.
If the paperwork says the intern is the CEO, then the intern is the CEO, regardless of how much the intern's agent's github says otherwise.
Yep. You're starting to catch on. Same functionality, significant cost savings for the organization.
This is hilarious, I feel like we are going to see AI companies add something to the system prompt or terms of service that explicitly forbids having an LLM act as CEO. Honestly though I do hope this gives pause to the C suite when it comes to replacing people with AI
We've replaced our CEO for the better part of 6 weeks now. He has been on a media tour promoting a product we no longer sell, but it keeps him busy while we continue to grow.
While Proxy AI [0] is not open source, it is free, it is built from pure garage energy, and we are having our summer sale right now. 20% off for HN users.
[0]: https://getproxyai.com/
I think AI CEOs are dumb but AI 'enterprise decision-makers' could be very very interesting.
Imagine how quickly your meetings could get resolved if you had teams submit prompts / contexts, had a clear, documented set of overarching objectives, etc.?
Let's also replace politicians and other officials.
AI - might finally "care for miners and not the minors on the island somewhere".
Therefore, AI will not be blackmailed into doing shit.
We're all joking around here. But 5 decades down the line, with increasing cognitive offload to AI, is a slow erosion leading to this end state so crazy?
Awesome!!!
This is the best project ever.
I have been waiting for this for awhile. Now we can have real layoffs for an AI pivot starting with management :-)
Many CEOs are bad and their businesses will either fail or be bailed out. But the work CEOs do is probably some of the last to get automated - higher order strategy, relationship building, managing people.
Honestly, this is hilarious!
Please replace CTOs while you are at it and most architects can also be replaced easily!
It can be configured to replace any mangagment role. The goal is to do for management what co-pilot, cursor and Claude did for development.
When will we see the world's first fully autonomous company?
If you want to see that idea extrapolated to an impressive extent, check out Accelerando by Charles Stross. If Stross was right we're in for a bumpy ride.
Is this automated CEO any good? Is there a listing from a demo session?
Highly configurable. Looks like it is targeted at any level of management. The key features seems that it can generate its own vision of how to move something forward and that it can be set with degrees of freedom depending on the use case. Video demo here: https://www.youtube.com/watch?v=O_g97xxVTMk
The speech model for the voiceover should be named “Max Vocalfry.”
> Is this automated CEO any good?
Most importantly, what is its golf handicap?
Peak comment right here
After firing all, when CEO can fire themselves to be just founders?
This will be epic.
Remember that guy who applied for CEO OpenAI and said that he will replace the CEO and the board with a chatbot
I wonder why OpenAI didn't accept it! It would he such a great proof that AI is soooo good if the frontier pab itself uses AI to replace execs
Everyone seems to be missing the point of the role of the CEO.
The CEO is not some masterful person answering all the questions for the company, I would argue that a CEO should be able to disappear for a year and if they had done their job correctly and assembled a good team the company would do just fine without them.
No the purpose of the CEO is to safeguard the Capital that has been invested in the company. That is all. They make sure it doesn't walk our the door in wages or other wasteful spending. They are the Capital's man on the ground.
Yes, this is actually a reason why an AI executive could be a formidable alternative in theory.
The CEO is ultimately an employee of the board, and managing the CEO with respect to the board's oversight and interest is a somewhat adversarial challenge.
However, the AI CEO's intentions are fully transparent to the board and entirely pliable to the board's demands. The AI CEO also offers the board the opportunity to retain significant equity shares that materially impact their wealth - in this way AI executives stand to be an even more tantalizing target than fixed labor costs.
Of course, this presumes LLMs will become reliable enough for this, which doesn't seem likely.
I would argue that many companies would in fact be better off disappearing for a year, rather than constantly thrashing the company to chase the latest trend.
And judging from the ZIRP hiring sprees and tokenmaxxing, CEOs are more often the driving force behind wasteful spending than they are the safeguards against it...
> should be able to disappear for a year
But I do my job properly and get in trouble when I ask for a week off
looks like a good way to run up a really large anthropic token bill.
Finally a use for Grok Bots!
But can it avoid answering direct questions by telling you about past challenges and what it's doing to address those next year?
What company fired them? I don't see that in the repo but maybe I missed it.
what's the backstory here?
Op here. Team was let go. Decided to do a startup. We thought that tools like co-pilot and Claude code were transforming development. We wanted to transform management (not just CEOs). We open sourced this knowing that if people find this useful (and we hope they do) we can get a support work out of it (similar to the Redhat model and other open source tech).
I’m all for productive people, brilliant engineers. I still get dirty with codes/work all the way down to the newly hired fresh out of college kids, to the production architects that shapes multi-million, if not billions, of P&L. I’m also very lucky to be included in key decision making, planning, negating amongst the boards of C-Suites, Directors, VPs.
A very personal view but many developers/engineers fail to see the overall picture. And yes, I was there and I went through all of the mistakes.
The things you see on the surface of those incompetent CxOs are just the tiny piece of the bigger picture. You cannot replace blames, decision in the face of uncertainty. Almost all decisions that businesses take have nothing to do with the proficiencies of technologies or engineering finesse, or even the best options, but mostly on “incentives” and who can take/own the “blame.”
“AI” replacing developers is evident because that is a rather tiny perspective and kinda easy to do the equation - developer can do that (good, bad, best) and can the AI do it - maybe, perhaps. So, replace.
A C-Suite took a decision to tick the final approval of a billion dollar deal with a service providers (amongst 10s of them). The idea might have sounded idiotic because there were 3 others better at the top than the chosen one. Even if all of the task and options were weighted with AI, but that last human decision is the Blame that the C-Suite took on, by betting his incentives on it (money, fame, pride, ego.)
> Almost all decisions that businesses take have nothing to do with the proficiencies of technologies or engineering finesse, or even the best options, but mostly on “incentives” and who can take/own the “blame.”
This argument always confuses me because it seems like lower level employees are the ones that get quickly dismissed when things get bad while a lot of C level executives seem to have some sort of compensation even if they fuck up.
I think a lot of this kind of talk treats the C-Suite as special, mystical "visionary" people who Know Things We Can't Know, and by virtue of this otherworldly magical power, are uniquely qualified to do their jobs. Like the average person with basic knowledge of his business and industry can't possibly choose one service provider out of 10 bids, or set business priorities, or drive and enforce culture. We're not talking about wizardry here.
People always dunk on me for saying this outrageous statement, but if you put me in charge of a random Fortune 500 company, I give myself a 50% probability of the company performing better than if its real CEO was in charge, and a 50% probability of it performing worse. I really don't think these guys bring as much to the table as they think they do and as much as the general public seems to think they do. Business success does not come from paranormal insight and abilities.
I think you could randomly nominate CEOs from the set of the general population with, say, a minimum of a college degree, and the randomly nominated CEO's performance would not statistically differ from "qualified" CEOs.
This is absolutely spot on. CEO/C-Suite/VPs are surrounded by gatekeepers (usually something like the PMO org, program management) who feed them with garbage info. Even if a top level leader is genuinely interested in making things better, they would never know because the gatekeepers firewall any competent person.
The gatekeeping org has the highest density of incompetence. They only hire incompetent because loyalty to the person hired is the most important. Gatekeeping orgs of various leaders protect each other, there is almost never any blame, it is things are good and they are going to get better, announce xyz initiatives/projects.
> A very personal view but many developers/engineers fail to see the overall picture. And yes, I was there and I went through all of the mistakes.
What is failling to see the overall picture? I hate this overall picture mysticism. I bet that if people where given the same role and information as the CEO they would be able to see the overall picture.
The difference is that the CEO controls the capital and has the ear of the investors, whereas the developers generally don't. It's the investors that control whether the CEO is a bot or not.
Honestly the best use of AI at the moment is to replace non-technical middle management layers. Many of those yelling the loudest for “more AI now” are quite ironically holding the jobs best replaced by the current creation of AI.
Number of people who think that a leader or a CEO can be replaced by AI and who themselves have actually risked everything to start a company, be a leader, or do anything similar: exactly 0.
What are your qualifications to say that?
> Number of people who think that a leader or a CEO can be replaced by AI and who themselves have actually risked everything to start a company, be a leader, or do anything similar: exactly 0.
Prove it.
> be a leader
There are a lot of bad "leaders" out there running Fortune 500s, fwiw.
If you just want a sociopath with faux accountability that's not very difficult, dime a dozen.
This has likely been the case as long as jobs have existed. Some CEOs spend their entire time tweeting, apparently.
The way any large company works, and this includes all the big tech companies, is that they end up in a state of constant churn. Every 6 months or so there's a reorganization. There'san old joke that you get an email that your manager's manager's manager's manager now reports to a new manager and you've never heard of any of these people.
The point of that churn is to make sure nobody is ever held responsible for something not working. It's never given enough time. That's by design. Middle (and even senior) management is about empire-building.
Another quote that springs to mind is that the bureaucracy is expanding to meet the needs of the expanding bureaucracy. That's how I feel about management bloat.
Anyone who has paid attention to Intel's downfall in the last decade will have heard stories about how the company descended into fiefdoms and it became impossible to get anything done. That's what happens.
Most people in these organizations aren't particularly talented at anything other than forming social relationships. the leadership of these companies always end up having an adversarial relationship with their workers and try to suppress their wages to increase profits so they can get their stock options and bonuses and the CEOs all sit on each other's boards of directors setting each other's compensation packages. It's a lootocracy.
The point of all this is that the leadership will never decide that they are the problem. Take Meta's Metaverse. It was a $70 billion flop. A lot of people got laid off from that. Were any of those worker bees responsible? Nope. They didn't choose the strategy or probably even what they worked on. It takes a lot for heads to roll within senior leadership and with the vote-stacked share structures of many major tech companies, ousting a CEO is essentially impossible.