AI is blurring the line between PMs and engineers?

I'm fortunate to work for a company with great PMs at the moment, but in the past, it's usually been the case that PMs blur the line between PMs and engineers by making us do their job for them. That they can use AI to appear productive is not a promising development.

Forget PMs and engineers, the discussion I want to see happening is just how much more easily an LLM could replace the average CEO...

> “No, no,” she said. “Engineers aren’t allowed to edit the prompts. It’s only the PMs and domain experts who do prompt engineering. They do it in a custom UI, and then the prompts are committed to the codebase.”

This feels like a desperate power grab.

Why can’t engineers be involved with the prompts? Why aren’t they allowed to do things like automated A/B testing or to implement ideas from papers they’ve read?

Banning engineers from prompts altogether feels extremely arbitrary. If this was a healthy relationship they’d at least say the engineers and PMs work together on prompts.

Banning them is just politics.

Wishing the best to my PM, who struggles to write a ticket title.

These company blog articles are usually for marketing. Humanloop develops software to assist with prompting during the development process, so the author’s conclusions reflect the company’s intentions more than an objective industry observation.

AI is transforming how we prototype and iterate, and products like v0 or Replit are scratching the surface. However, historically, low-code platforms lacked a good integration with complex development cycles. There were many attempts, but they either failed or shifted their focus: Microsoft Expression Blend had a brilliant concept of integrating early sketching and ideation with development, but the product ultimately died with Silverlight; Framer had an editor that allowed users to integrate React components with a design tool, but they repurposed their product into a CMS-oriented tool like Webflow; Builder.io is following a similar path. It seems that in today’s market, there is no clear fit for the RAD tools of the late 1990s. Maybe AI can change that and create the new equivalent to Visual Basic. The hardest part is the extra mile that goes from the prototype to something robust and complies with multiple quality attributes: scalability, performance, security, maintainability, and testability.

I’m SWE on a product and generally agree with this, at least for user-facing prompts. Once you’re walking an LLM thru OOXML it’s better done directly by ENG.

For tools, not clear how this works since as you adjust parameters and whatnot you’re also presumably changing the code downstream when you “execute the call”.

But probably both sides of this will be done by LLMs directly in the future. I rarely write or tune prompts by hand now.

A joke: Good PMs will do invisible work. Bad PMs are just invisible.

I've been watching several "vibe coding" vids on the youtubes. https://x.com/karpathy/status/1886192184808149383

Some of them are painful. Some are impressive. The projects are small. Sometimes pulling in powerful off-the-shelf modules. They are getting better fast.

As a greybeard software architect, my current annoyance is that I'm spending all day talking to people when I want to get some practice voice prompting code :P

i think the plan of folks like these is keep peddling agi, replacement, line blurring and make the money through secondaries for as long as possible till it all collapses like crypto.

PMs (generally) can't tell if AI wrote bad code.

Absolutely agree with this observation. And I think it’s a great outcome.

The vibe coding Reddit (http://reddit.com/r/vibecoding) already contains the full spectrum of “first time trying to code” to “just rolled my own custom GPT to optimize this.”

These articles are a little annoying.

I love ChatGPT [1]. I use it all the time. I use it for coding, I use it to generate stuff like form letters, I use it for parsing out information from PDFs. Point is, I'm not a luddite with this stuff; I'm perfectly happy to play with and use new tech, including but not limited to AI.

Which makes me confident when I say this: Anyone who thinks that AI in its current state is "blurring the line between PMs and Engineers" doesn't know what they are talking about. ChatGPT is definitely very useful, but it's nowhere near a replacement for an engineer.

ChatGPT is really only useful if you already kind of know what you want. Like, if I asked it "I have a table with the columns name (a string), age (an integer), location (string), can you write me an upsert statement for Postgres for the values 'tom', 34, 'new york'?". This will likely give you exactly what you want, will give you the proper "ON CONFLICT" command, and it's cool and useful.

If I ask it "I want to put a value into a table. I also want to make sure that if there's a value in there, we don't just put the value in there, but instead we get the value, update it, and then put the new value back in", it's not as guaranteed to be correct. It might give you the upsert command, but it also might fetch the value from the database, check if it exists, and if it doesn't do an "insert" and if it does do an "update", which is likely incorrect because you risk race conditions.

My point is, the first example required knowing what an upsert is, and how to word it in a technical and precise way.

It certainly doesn't "blur the line" between PM and engineer for me. I have to pretty heavily modify and babysit its outputs, even when it is giving me useful stuff. You might be saying "well that's what a PM does!!", but not really; project managers aren't typically involved in the technical minutia of a project in my experience, they're not going to correct me for using the wrong kind of upsert.

These kinds of articles always seem to be operating on a theoretical "what if AIs could do this??" plane of existence.

[1] Deepseek is cool too, what I'm saying applies to that as well.

ETA:

Even if I wasn't a fan, this article definitely shouldn't have been flagged.