Really should add a poll system for Forem... Might do that this weekend.
A bit of background, for the past year and a half, I've been working as a...
For further actions, you may consider blocking this person and/or reporting abuse
It reminds me of my own year. My skill got reduced to an OS choice on a laptop. Yours got reduced to a subscription price. Neither one was ever about the actual work. 27 minutes for full severance is a clean way to end that story.
That's messed up! Yeah, the real thing that just pisses me off, is if I am 'redundant', dont try and boot me through conceded methods to justify it. Shoot straight, tell me 'listen, we're downsizing soon, unfortunately we wont require your services anymore, I know notice period is 1 month, but we're telling you in advance, we have runway for 6 months for you, after that it's uncertain, please look for alternate employment, so we dont leave you stranded. We appreciate your work, but we simply dont need you anymore'. That would have immediately given me a deadline, clear intent and I wouldnt have the stress and anxiety of dealing with all the 'complaints' that were just a fabricated paper trail for them to pull the trigger... Honesty, transparency, clarity, trust...
A clear deadline beats a slow dismissal. They chose the fabricated trail because it covers them legally. Honesty exposes more risk than lies do...
Yip... That's the sad part and exactly why 'meeting with management' was said instead of 'come have a meeting with me and my lawyer'...
You know. I sent out probably 1000+ CVs, gotten back maybe 20 replies (sorry, we went with a different candidate), worked my ass off building projects, it absorbed my weekends, it absorbed my 5-12, it absorbed my life. In the end, I wont say that I got the Wasmer interview out of sheer talent, I'm sure plenty of people could do what I did. Hell I'd admit it, I was pumping out projects so fast, that I dont know how half of them work, because I literally dont have the mental capacity to be at 12 places at once. In the end, volume didnt get me in, quality didnt get me in, it was sheer dumb luck... Now it's on me to not screw it up during tech eval.
My backup plan though, was to go solo, write a single app I'm proud of, release it. I had settled on a health and fitness app for my fiance. I'm still gunna do it regardless, but that was my backup plan. Your skillset has shifted entirely into AI, USE IT! But dont use it as a service, use it as a product. Nigeria is a big place and you have the skill to make apps that solve real local problems. I had always thought of how GOG operates, 'just sell a copy', that's not gunna work today... 'sell a subscription', why? People just clone it with AI in a day... 'use ads', people see an ad, delete it. Inference... Adding a 10% markup on llm API fees, you wont make a ton of money, but it'll be consistent. Keep the app entirely free, BYOK too, most people dont bother and pay the 10%. It's consistency, that's why anthropic, openAI, Google, even eastern companies like moonshot, make bank today. A small markup on service delivered, but you know the income is consistent and growing.
Unfortunately, the main AI features are already claimed by large companies, but that created it's own window! Take a health and fitness app, so many exist, all of them have ads, or charge a subscription, or paywall features, everything is 'pay, or get a lackluster service'... Free and BYOK, make paying you optional, that transparency sells!
Something that I've been thinking about alot lately, there's so few people who actually know enough about AI, that they can tell you Claude isnt always better than Gemini and there are times where Kimi K2.7 STILL wipes the floor with any other model from any other provider. Make a list of all the Cloudflare workers AI models, then find their niche. Each is exceptional for their price point for something? So build a little cloudflare function, that acts as a model router, keep it simple 'high, medium, low', charge a flat markup rate, eg. 10%, then host it as an OpenAI compatible API. I sure as hell would rather pay a 10% markup and have the API automatically delegate a task down to gemini flash lite when it's simple enough and scale it to claude opus only when absolutely necessary... And that's the real market gap at the moment. People pay extortion level prices for simple tasks, because they know too little to actually swap models for lighter tasks. It'll maybe take you a day or 2 to set up properly and wont cost you a dime while it's idle. Marketing for it is simple, set a scoped task, execute it with a flat model and then via your API, compare the time to completion and the cost. People who rely on Sonnet medium get their job done on their subscription, but it takes 30 minutes to do what gemini flash does in 10 and what qwen does in 5... and even at API pricing, 25 minutes difference is real work getting done, instead of idle time.
Give it a go! I know an idea to run with isnt much, but I hope it helps. I know it's a nasty world out there trying to get paid... I hope you go through with this 1 though, even if it's just another notch on your cane, I do believe it's one that can help pay the bills.
The model router idea is solid. I already build on Cloudflare Workers AI.
A high, medium, low router fits directly into that stack. BYOK plus a flat markup removes the trust problem.
People see the real cost instead of guessing. I could have a working version running in a weekend. Thanks
Man, that "$20 Claude subscription" comment is an absolute joke.
People need to realize that an LLM is just a tool, not magic. Giving a raw $20 prompt-bot to office devs who don't even write their own commits is a fast track to disaster.
Without deterministic memory, AST context, and proper orchestration, LLMs don't "understand" engineering logic. The moment a non-trivial bug hits or a feature requires deep structural changes, a raw chat interface gets stuck in a hallucination loop. It will just start blindly rewriting the entire codebase and breaking existing logic because it has zero long-term context.
A prompt wrapper doesn't have 6 years of architecture decisions in its head, and it sure as hell won't save them when the ERP falls apart. The stale build-cache warning is just a cheap excuse from management to hide their own incompetence.
Let him click his $20 buttons for a month or two. When the system breaks and their "AI-powered" team has no clue how to trace the root cause, watch how fast reality hits him.
You’ve got 5 months of runway, your own product (Doccit), and Wasmer lined up. Protect your health and your back — hand in that resignation and let them test their $20 theory in production.
Good luck with Wasmer!
Thanks! Yeah fingers crossed on Wasmer, it'll all pan out regardless of if Wasmer works out or not. Kinda why I waited till I had the runway to guarantee we'll be fine before I quit, cuz even if I get a new job, probation period is still a thing and I'm only safe once I've cleared that. But I should do fine I think.
As for my current employer, it's going to be very satisfying tomorrow to tell him I quit and hand over the paper... He'll have exactly 1 month of my time, then I'm out. He needs to learn, if you pay a developer pennies, while screwing them over... At some point they're gunna reach the point where enough is enough and walk out...
Bro, just wait. Give it a couple of months, and he will be calling you to come back for whatever price you name.
But the funniest part will be watching the next dev try to take over your core systems. They won’t understand your architectural decisions, they’ll start rewriting things their own way, and inevitably break the system somewhere else in the process. Enjoy the countdown tomorrow!
He mandated, that the manufacturing system be built using CSLA, if you've ever dealt with that... Then you'd really feel why I created V.A.L.I.D. But it wasnt allowed to use V.A.L.I.D. for it, so instead, they'll have to learn CSLA and try not to break anything (practically everything is intertwined 🤣 You cant use AI in there, because it cant comprehend more than a 3 tier linkage, that system has 18 layers of interops, each entwined with the next and previous, so 1 action cascades into more. Cuz it's formulas, works orders, migrations, stock tracking, ledger entries, double entry accounting, not to mention it does that internally, then it needs to still translate to Sage Evolution and interact via it's API properly... THere's no way they'll wrap their heads around that 😁
Honestly, that final month is going to be absolute hell. They’re going to ride you hard for every single breath, micromanage every minute, and try to pin every issue on you just to push your buttons. Don't take the bait, and definitely don't waste your energy trying to teach them anything or leave detailed guides — actual knowledge transfer costs real consulting money. Just protect your health, do the bare absolute minimum, and let them play their petty games while you count down the days.
That's the plan. From here on out, I am on minimal effort. I do my job at a reasonable pace, with reasonable quality, taking it nice and relaxed. No running left right and center to fix it all up, no detailed guides or anything. I do my job, till my last day, then I'm out. No aftercare support, nothing. They can figure it out themselves... If I was treated fairly, I'd have been more than willing to help out and transition myself out cleanly. But after all this... They can shove it, they're getting no sympathy from me.
The most striking part of this post isn’t really the AI argument—it’s the breakdown of trust underneath it. Once an employer starts questioning remote work, medical circumstances, code quality, and AI usage in the same period, every individual disagreement becomes part of a much larger workplace dynamic.
The “I hired a person, not an AI” comment is especially interesting because it frames AI assistance as a replacement for expertise rather than a tool used by someone who already has expertise. There’s an important distinction between relying on generated output blindly and using AI as part of an experienced developer’s workflow, with the developer still responsible for architecture, verification, debugging, and decisions.
The code-quality incident also highlights why process matters. A stale build cache leading to a bad push is a technical failure worth addressing, but attributing responsibility requires looking at the entire development and testing workflow—not simply attaching a warning to the person whose commit happened to expose the problem. Good engineering organizations turn these incidents into process improvements rather than ammunition.
The remote-work conflict is arguably even more significant. Whatever an employer’s legitimate business requirements may be, combining pressure around office attendance with an employee’s medical situation creates a situation where documentation, clear policies, and appropriate boundaries become extremely important. The request to hold a management meeting at the employee’s home also raises a reasonable question about professional boundaries when remote alternatives exist.
I also think the financial runway section is one of the most useful parts of the post. Five months of savings, a second income stream, an existing product, independent-business experience, and active job applications dramatically change the decision from “Can I afford to quit?” to “What risks am I willing to accept while transitioning?” That’s a much more concrete framework than making the decision purely from anger.
The one part I’d be cautious about is the theory regarding why management is behaving this way. It may be correct, but the post itself doesn’t establish that motivation. The observable facts—changes in remote-work expectations, conflict over AI use, the warning, and management’s requested meeting—are enough to evaluate the situation without needing to settle the question of intent.
And there’s an important distinction between leaving professionally and leaving dramatically. If resignation is ultimately the choice, preserving documentation, understanding contractual obligations, protecting finances, and keeping the exit factual may serve the author better than getting the final word about Claude. The strongest leverage is often simply having options.
The bigger question underneath “to quit or not to quit” seems to be: how much energy should you spend trying to repair a workplace relationship after the underlying trust has already deteriorated? Sometimes the answer becomes clearer once you stop asking whether you can tolerate the situation and start asking whether the situation is worth continuing to invest in.
I had the meeting earlier today, it went well. Before the meeting, I gave him a heads-up, that I dont think my back will be in a state for a 2h commute both ways and an 8h office job anytime soon. They clearly need someone to sit at a desk so they can supervise, I cant do that. So I suggested a MSA, so I can focus on recovery before I start anything new. Went to the meeting, was boss + a labor lawyer, 27 minutes later, the paper was signed. I get my full severance package (1 months pay + accumulated leave + 1 year's severance) and my employment is terminated effective immediately.
I saw in the lawyer's stack of docs, that they had prepared a SLA, so chances are that he wanted to out me as an employee and instead take me on as a contractor for remote work, which would have meant a considerable pay cut and non-guaranteed monthly income. So overall, I think the outcome was spot on for how I would have liked it to go.
Dont get me wrong, I really loved working with the team, seriously talented people and kind, but only 1 of them seemed to have any proper procedural communication skills... Everyone runs around looking over shoulders and quick comments and think that can replace well scoped work. The 1 guy though (the 2nd in charge), brilliant, absolutely brilliant, I loved working directly under him, because he'd write me a solid scope, not just expected capabilities, but necessary constraints, when he gave a task, I had it done in no time and if ever I needed to adjust it, he'd give me a surgical scoped fix I need to implement, not just 'yeah, I think we can separate these panels', more like 'can you separate these panels by sku type, that way it's a bit easier to manage and would make the excel export cleaner'. That's the difference between someone who can work with remote workers and someone who cant. And it shows, because he's the only 1 that was there during Covid times, so he learned the hard way how to work remotely. The rest were born and bred in an office.
The 18-layer interop system is a beast. No cheap AI can untangle that. Wish you a smooth handover!💪
Thanks, yeah it was a pain to integrate, especially the cross-validation and recursive quantity lookup... They're gunna have fun maintaining it 🤣 Even Fable at ubercode couldnt figure out how it works, so they've got their work cut out for them
So I went to the meeting. Was boss + labor lawyer. We signed a MSA, full severance, effective immediately, so that was awesome! I saw in the lawyer's set of docs, he had prepped a SLA, so I think they were going to negotiate that I work remotely, but at reduced pay... So overall got exactly what I wanted and without even needing to fight for it 😂
Wow that’s incredible news! Huge congratulations! So glad you got full severance exactly as you hoped, what a great outcome 🎉
Right? Honestly, I had expected a struggle, but it all went smoothly and literally 27 minutes (including walking from and to my car), we were done, signed and I'm clear 🥳 Now I can focus on the Wasmer tech interview and then have an entire month to relax and recover till I need to work again 😁 So It's awesome!
That’s such a clean win! 27 minutes is insane. Good luck on your Wasmer interview, enjoy your well-deserved month off 🎉
I submitted a comment, and it disappeared? This is the 2nd time this has happened to me.
It's that new filter they added to try and get rid of bots... Unfortunately it's broken and it flags practically any comment with a link in it, but they are fixing it, I already reported the bug a week ago with some thoughts on what's causing it, so they are actively working on it.
Consutar las leyes sobre Derecho Laboral de tu pais, y valorar. Animo. Gracias por el post.
Already did, they are being very particular with how they've been phrasing things, so they dont trigger any red flags for labor law.