Tyler Postle

@iamireneyu And theres nothing wrong with using those tools, theyre great! I just get frustrated when ppl claim theyre perfect or theres no value in understanding fundamentals, the classic “youre cooked” claim lol

@PostleTyler it's very true.. having used a bunch of these tools including lovable and claude code myself, let's just say it helps a lot to understand things like "session storage" and "state persistance" and be able to direct the AI to well.. set up an app correctly.

A PM spent a year learning AI (LLMs, RAG, agentic workflows) then realized she still couldn't do her job. The problem wasn't AI. It was everything underneath it. She'd been pushed into AI product development at her company. She did all the right things: took courses on generative AI, learned about LLMs and RAG, studied agentic workflows and orchestration patterns. But when she got into the actual work, something felt off. "I still feel like if I don't have that non-AI product foundation, I can't really do this," she told me verbatim. "Because those systems aren't going anywhere, AI just expands on it." She was right. Here's what most PMs miss: 👉AI is still software.👈 It still runs on servers. It still uses APIs. It still stores data in databases. It still has latency, architecture, and infrastructure constraints. When you don't understand modern cloud-native systems like how requests flow through APIs, how data gets stored and retrieved, how servers handle load, you can't lead AI products effectively because AI fluency requires systems fluency as the foundation. The client-server model (presentation, business logic, data layers) doesn't disappear with AI. The AI model *is* the business logic layer – just a more complex version. It still needs APIs to connect to the frontend. It still needs a database (often a vector database, but still a database). It still lives on servers that have to handle requests. If you don't understand how those pieces fit together in traditional software, you won't understand how they fit together in AI software. This PM made the right call: she stepped back from the AI hype and invested in learning the fundamentals. Not because AI doesn't matter, but because *you can't build on top of a foundation you don't have.* The irony is most developers building "AI products" aren't building AI models either. They're integrating existing models from libraries like Hugging Face, wrapping them in software architecture, and connecting them via APIs. The work is still fundamentally about systems: how things connect, how data flows, how to architect for scale and reliability. So if you're feeling anxious about AI, ask yourself: am I actually behind on AI, or am I behind on the technical fundamentals that make AI possible? The gap you're feeling might not be what you think it is. What technical concept do you wish you understood better (AI-related or otherwise)? --- If this was useful, share it with a PM who needs it. I teach PMs technical leadership every week. Follow along so you don't miss the next one. Want a head start? Free "Tech for PMs" mini-course: skiplevel.co/mini-course

Boardy's been in production for years across multiple channels and millions of conversations. Andrew's honest take on evals: if your user experience is well-defined, evals are straightforward. If you're trying to measure something like long-term goodwill, it gets a lot harder. Most teams haven't hit that wall yet. They will. @andrewdsouza @boardyai

@SergioRocks A good manager is so much more than a progress update or a cog in a machine. Its good that those kind of management jobs arent going to hold as much weight. Makes way for real leaders, coaches, team players.

The first jobs silently killed by AI are middle managers. People whose job is managing process are in less demand than ever. For years, teams scaled like this: - More people → more complexity → more managers. Layers of process just to keep things aligned: - “Can you share a quick update?” - “Let me sync with the other team” - “We need approval before moving forward” - “Let’s set up a meeting to discuss” That made sense when the output of doers was the bottleneck. But AI changes that. When one person can do the work of 3, you don’t just reduce headcount. You reduce the need for coordination. - Fewer people → fewer handoffs → fewer layers to manager. The middle management erodes, and we get flatter teams where: - Business stakeholders bring scope clarity - Builders own end-to-end features - AI handles parts of execution that used to require more people This is not about replacing teams. It’s about reshaping them. The companies winning right are removing friction. And most of that friction was sitting in managing a large group of people.

@hthieblot I never could have imagined the weight and pressure that this opportunity and responsibility comes with. Its heavy, but its also fulfilling

Raising money isn’t the hard part of being a founder. Here’s what actually breaks people ↓

@molecularmusing You’re not alone. But it does feel like people like us are few and far between these days. (Maybe its just this echo chamber)

I find this extremely worrying, with many of people I respect saying things like "I no longer write code" or "let LLMs do it". Why did you start programming? Was it never the journey for you, but only the goal? I genuinely want to understand this, I seem to be the odd one out.

@andrewchen I mean more than that, its just that humans will always have economic value and our system compels us to work more with the extra time we get freed up

Thus, the post AGI world won’t be a life of leisure where we frolic in nature 24x7 and stop working Because the most fun thing to do will be to build cool new stuff using AGI 24x7. We’ll be glued to our computing devices as a result 😂

@garrytan Hehehe this is rich coming from the company thats promised hundreds of agents, and is trying to acquire their way into that promise. Thats all I can say on that at the moment

Recent earnings call, Aneel Bhusri of Workday says startups with AI agents are "parasites" This is what system of record incumbents really think of startups. The war is just beginning. The facts: the user data belongs to the users, not the incumbent software vendor.

@garrytan Wont be close to the real thing, but there are a million little things you need to know to set your startup for success and YC gives a great foundation across all the content they share, so probably worth checking out

I just launched /office-hours skill with gstack. Working on a new idea? GStack will help you think about it the way we do at YC. (It's only a 10% strength version of what a real YC partner can do for you, but I assure you that is quite powerful as it is.)

@chrispisarski @crustdata Very interesting! Any tips on GEO? What tools do you use to monitor and improve it? (Working on improving ours as we speak)

ChatGPT went from 43% of our AI-sourced qualified leads in September to 3% in March. Claude went from 4% to 43%. Nearly all of that change happened in the last 8 weeks (!!) Even though @crustdata gets 2.5× more sales demos per month from AI than 6 months ago, the share coming from ChatGPT has collapsed Probably one of the biggest shifts so far: buyers with intent are now using Claude for discovery more than ChatGPT

@chrija I mean that’s better than agent-agent ecosystem pitches

"Context-aware orchestration layer between humans and AI agents" If I hear this pitch one more time, I can’t guarantee anything.

@itsCathyDi I like it! A model that can take action. I put a few more “advanced” components to agents in my definition here, but I think the essence is the same. linkedin.com/pulse/prompt-e…

we cannot build what we cannot picture we cannot picture what we cannot define agents := model(s) + tool(s) change my mind.

Theres still so much that can be done to improve agents with all sorts of optimization, analytics, API standards , etc. but yeah I wouldnt bet on models becoming god This is what happens when you have to outcompete your own claims. Time to roll up your sleeves to improve your agents

Altman admitted that transformer models have hit the wall. Most improvements in the last 9 months are attributable more to the tooling around the model rather than the models themselves. In other words, this technology is rapidly maturing with no signs of another leap.

@Alex_TheAnalyst These ppl already existed, but I agree itll only make the problem worse. We need to continue to preach the importance of critical thinking. The WallE prediction wasnt about people not exercising it was about losing the will and ability to use your brain

I'm going to call this right now. We are going to have a large population with absolutely no critical thinking skills if they blindly trust AI for everything. We have all already seen it. They don't validate outputs. They don't really understand anything. They just ask questions, it looks good, and they go with it. There are going to be huge issues in every company as this continues over the years. The amount of technical debt and knowledge gaps are going to be insane. So much opportunity if you actually know what you're doing.

@DrDominicNg Also can look to @DOTA2 for a case study. My first intro to OpenAI was watching their AI play against international champions. Pro scene still exists, AI hasnt taken over

Chess is 30 years ahead of every other profession in dealing with AI. The best case study we have for what's coming. 4 lessons: 1. Human-AI collaboration had a 15-year shelf life in chess. "Human in the loop" is a phase.

@tszzl If automating jobs meant people would have to work less and make more (with 1:1 correlation), people would be all for it

one of the most profound cultural differences between san francisco tech and elsewhere is that tech people see automation as axiomatically a good thing, and this idea predates ai by decades. most of a tech company is automating oneself out of a role by writing software or hiring

@conor_ai Nah, lets just play some @DOTA2

claude is down do we… go outside?

@iamireneyu Wrote an article on this linkedin.com/pulse/prompt-e…

@iamireneyu THANK YOU. So true. Don’t eat the vibecode propaganda. Real knowledge still matters. Prompt engineering was a hypeflop because agents are so much more than prompts. Turns out building agents looks a lot like building software. Who’d have thought!

@johnloeber This just in- yc closes all future batch applications. After the wild success of gstack, yc realized they only need one person to build the next 1000 $100B companies Founders are now part of the permanent underclass

you’re pitching garry tan “so what do you guys do” you start explaining he’s furiously typing . two keyboards. one hand on each. you’ve never seen this before “who are your top customers” you explain. he types. his apple watch is a strobe light of notifications “who’s your competition and why should i invest” you explain that there’s no competition and you are the best and only product in the space “false!” garry jumps out of his seat “i am the competition!” you are speechless “in this meeting, i vibe coded your entire company. and my gstack has already closed your top customers.” you check your phone. your stripe graph shows 100% churn “and look at this” garry shows you his imessage. there’s a text from 35 seconds ago. your top enterprise prospect that you’re trying to close? garry’s AI is trading baking recipes with the CEO’s mom “thank you for playing!” you have no moat. you are not admitted to the YC spring 26 batch.

@jonnoxrevanche Reading, learning, true critical thinking will be just as- even more- important than ever. Those who spend the time and effort to truly learn will have an even greater advantage

What chatgpt has revealed is that a surprising large number of people want to be “seen” as intelligent, profound or insightful. Many more people than you’d assume. But they profoundly resent intellectual work or reading or study or, say, literally anything you do to get smarter