April, 2025
CXOs of a few companies have asked me if I could teach their devs to use AI for code. They should ask me to cure cancer instead. If a university degree only taught us one thing it should be that when you force people to learn something, they do everything except learn that thing. I can’t make people learn things that they have no interest or benefit in learning. AI might increase their productivity by 10x but they are still being paid for a 9-to-5 job and they are working the same time after AI that they were working before AI. They gain nothing from trying to keep up to date with the latest in AI when everything in AI changes every 2 weeks. The top down approach works as well as the SMS OTP services in Pakistan. But what if an engineer working in a company wanted to do it bottom up and they really pushed for it. Well, they should expect to be rewarded as well as Semmelweis. Semmelweis was a doctor who realised that less people died in his wards when he washed hands. So he pushed his colleagues to start washing their hands before procedures and they rewarded him by forcefully committing him to a mental hospital where he was beaten by the guards to death. The existing system of every company is balanced in the worst meaning of the word. It might be a balance of bosses who don’t understand things they manage; or executives who only care that people listen to the new meanings of the company vision; or a balance of the curious workflow of PMs with their daily stand ups to listen to explanations that they don’t understand and update Jira boards with a ticket for every time an engineer goes to the toilet so they can keep the CEO in the loop on how nothing is on schedule. So in short, you can’t mess with anything bottom up because messing with one thing will make the whole system of dumb processes go out of whack and with them something important might also likely stop working and everyone will have your eagerness to blame. As an engineer you even have your own team to push against which is an exceptionally bad idea because when you annoy people who lack the social skills to handle conflict, they just throw you under the bus. And do you think if you’re using AI your 2 story points are going to be accepted by the senior engineer who says it will take 10 burger points. You don’t hold the cards and your team just wants to play poker (they just call it planning poker so people can’t tell they’re just playing poker and not working). But what if you try to align both the top and bottom on the implementation. There will be humans living on Mars before you make this happen. The primary job of every system in a company is maintenance where they need to make sure that everything that is working keeps on working as they make small changes. This might not seem like a big task but this is the most difficult of all tasks. A company is a collection of debts. Debts in every sense of the word other than monetary. Every policy is a debt that needs to be followed by every new change and decision. Every existing functionality is something that needs to work with every new functionality. These debts make it impossible to make major changes. Every major change is like starting the company from scratch. In fact it is worse than starting a new company because of all the debt that you have collected. Even trying to reallocate some people to AI implementation won’t work because it does not work if only parts of it are updated. You can’t put a new engine on an old airplane. Even if it fits, the plane can’t bear the speeds that the new engine can take it to. The dilemma with the ship of Theseus is not whether it’s the same ship as you change parts of it but whether it works while you change them. If you paused maintenance mode, the company would collapse immediately. But if you don’t implement AI, the company still collapses eventually. This is why big companies die and get replaced by smaller companies doing some parts better. Adobe got replaced for product design by Figma and Figma is now getting replaced by people designing directly in code (which is why Figma is also building an app builder) So big companies are doomed. Kind of. There might still be one way out but it is the hard way out and you don’t expect leadership of big companies to rock the boat with their risk aversion, fuel cards and pension schemes and all that luxury to keep telling people to “get on it” without adding much value. Leaders are fine if the ship capsizes if they don’t risk getting blame for it. This is why big companies die of gradual irrelevance instead of sudden attacks. Blockbuster used to be a big company selling DVDs and they could not risk doing online movies so they lost to Netflix, gradually at first and then all at once. Big companies (and some extremely dumb small companies) are possessed by demons of KPIs and OKRs. These demons hold off bigger demons of investors and board members. But you can’t make KPIs or OKRs around something that requires creativity. What would 20% more creativity this quarter even look like. So in their obsession of measuring everything they lost sight of all things that were important but could not be measured. The famous quote “What can’t be measure can’t be improved” had a condition that said “not everything that can be measured is important”. But these numbers people are rarely thinking people. In fact the reason they convert everything to numbers is so they don’t have to think and can tell others and themselves that they did something because the numbers told them. They missed not the forrest for the trees but reality for numbers. But that is fine, they always have the economy to blame. But what if they just ignore everything that is happening in AI and just hope they don’t get affected. This works for technologies that are limited in their real world applications like blockchain. People who wanted to see how they could use blockchain learned that their only option was to run coin offering scams. Those that learned to use these scams made a lot of money but those who ignored them did not lose money. It is different for AI. The applications of AI include everything. AI is already in your personal lives whether you invited it or not. And it is going to be in your business whether you implement it or not. The only question is whether your company will be the one that survives with AI or someone else’s. If you’re starting to panic at this point, AI has you exactly where it does not care for you to be. You could perish for all it cares. But fortunately for you, I care (it’s more of a contemporary obsession than care if I am being exact) about this enough to provide a rough guideline to avoid perishing. If you’re not panicking at this point, sorry to hear that your company is going to go bust in 5 years. The difference is too big for you to escape its aftermath. That should be a cause of concern or excitement depending on whether you see the glass half poisoned or half breath of fresh air. Now the solution. You need a parallel company unburdened by the normal company debt and BS as you switch over to the new way of doing things using AI. I can share an example of the difference this parallel company makes. One of the companies that I am working for has replaced their entire engineering team with a single person and they are getting more shipped with less bugs. The single person costs a lot of money but it is still cheaper than an entire team especially considering how much less time it takes to ship(higher rates but lower total costs). Another example is automating the manual workflows of a fintech that took 3-5 days down to 3-5 minutes in less than a month. The longest time in this automation is the integration with the existing team. If the existing company tried building the same flows, it would've taken more than 6 months and they would still not work. While you need a parallel AI company badly, the overlords at OpenAI and Anthropic have decided that you must suffer before you get entry to AI and automation heaven. The recipe for the 2nd parallel company is one cracked engineer mixed with understanding of both business and product. The ingredients are rare but the bigger problem is most of you have Covid and your taste buds to identify these ingredients don’t work. You can’t tell when someone is super smart or when they just know a lot of stuff. So your first problem is finding these people. After you find them there is another problem. These people do not want to work for you. I say this from experience trying to convince these people to work for these companies. A university student who I had a company paying a 200k salary for a part-time role terminated his contract when I terminated mine. (I terminated mine because service based companies pay you for doing nothing other than taking meetings and I genuinely don’t want to read messages from the VP of a bank saying something is not working when his screenshot shows he is not connected to the internet). These people only want to work with specific people. When I first read the stories of how Steve Jobs would spend two years trying to get someone to join Apple, I thought it was so overboard and he could just get people who were really interested in the company. But the reality is that is that no one is actually interested in your company until they start working in it. Before joining they’re just interested in the image you have with your fancy offices and annual trips or they are just good at writing cover letters feigning interest. Do you seriously think that applicants at P&G are interested in making soap or it was their childhood dream to build b2b SaaS. So you might not be able to get these people directly and might have to rely on outside help. If you’re an engineer in the company you need to convince the leadership to not only create this parallel company but also learn from it enough yourself that you can move to it because the end goal of the parallel company is to kill the old one by cannibalism. This is better than being killed through torture by outsiders because it might allow enough time for some to transition. Because if you don’t learn enough to transition you will need to look for a new job in a world that you no longer understand. Don’t keep this problem of pending doom to yourself. Go and wake others and make them see the writing on the wall with the gutka spit on it. Your boss and CEO scare you often, send them this with the message “what are we doing about this”. Get your thin and lean AI company now from the nearest AI pharmacy. If the situation is extremely bad and your company has developed chronic constipation and is not able to get any new features out, consult an AI doctor.
---If you want to be notified of new articles, this is my substack where I send out unscheduled emails containing new essays and things that I am thinking or reading.
In case you need to reach out through a mutual or want to see my era of shitposting, this is my linkedin. Though I don't open my linkedin anymore