I’m tired of these AI written articles man. There could be information worth sharing here but having it all generated makes me assume you don’t care. So neither do I.
I'm working on something similar called devtop. You run it via docker and it has a UI for viewing artifacts and talking to an LLM.
The default workflow is 1) you create or derive documentation from the source code 2) PRD is derived from docs 3) tickets are derived from PRD 4) agents can work on tickets. All artifacts are files which live in the repo.
I like the idea of using claude code in a separate git-managed repo. I do this for the management side of my job, and use Obsidian as the human interface on top.
However I would never use someone else's starter repo. The whole point is to augment my own perspectives and strengths. On top of that, the AI written pitch is extremely off-putting. If you can't even bother to take the time to write your own pitch, why would I believe for a second that your workflow has anything unique that goes beyond the lowest-common denominator AI capabilities.
I transitioned into a Product role this year, coming from a CS/SWE/DS background and it is just shocking how poorly the AI models perform at written content.
Claude.ai Web (non-Code) has this feature where you can write into /areas/ or /topics/ and these files are automatically updated in other chat's sandboxes (but not re-loaded in those chats).
You cannot write the files directly nor can you download them directly, but you can ask a chat to show you the content or only the mermaid diagram that exists in workspace.md (or however you want to name it), and mermaid can represent quite complex networks out of the box, their relations, what is blocking something else, and so on. Chats can then edit this file with search-and-replace instead of re-writing it all over again.
Any product manager worth their salt should be in a coding interface--the line between Product and Engineering is blurring a lot more now, and to stay competitive PMs will need to reach a little further into the technicalities. Otherwise you risk just being another “idea guy” pushing tickets around, basically customer support on steroids which is getting eaten by Agents. Connecting the ideas to something tangible and translatable that an Agent could not do themselves is the new game.
(I am a PM of 15 years trying to navigate all of this…)
Honestly, the last thing companies need is more implementers. We had this before with UX prototypers. Focus on the what, not the how.
I'm seriously worried that my and other companies in my network have product leadership that are wasting their mental bandwidth on AI fun stuff, needlessly writing their own random tools, etc - while the product scope is going in random directions with zero product vision. This is not about job protection or "stay in your lane".
Lucid product vision is kind of the bar for “good product management”, I agree that slopping on top of a lack of direction is just going to make things worse. The issue with PMs going off on tangents with AI is the lack of engineering know how that they’ll need to adopt to stay competitive.
Sometimes I wonder what 'product-level' code really means. What actually is a product?
The code that goes into PLC equipmentss and hardware is incredibly simple—but it's still a product. If you look under the hood of a lot of software, the code is a mess. Still, it's a product. And when Claude's code leaked, everyone laughed at the code hygiene—but a lot of people still used it.
I belief is product simply has come to mean the most visible layer of the thing. For hardware "visible" layer is a literal device but more important product is what the device can do, a port is useless if there are no driver codes for it. For websites product can just mean the site/app does the thing, regardless of if it does it well, consistently, or could be done better. This is why words like "taste is what matters" and "code was never the goal" are said so often, I argue code is in the point but more people just care it works without going much deeper.
A product is what people (or now agents) can use to carry out any activity.
There is a lot of hand wringing in the software space about ai code being a mess - but from a product perspective code quality doesn’t matter. It’s the reliability of what people (or agents) use that matters.
Code quality used to further that goal but only because it was really hard to rewrite or fix code. If that barrier is removed does anyone really care what the code quality under the hood is?
Personally both apply to me, I also don't care about projects where the author hardly did any coding themselves.
The only semi impressive vibe coded projects I've seen so far have been rewrites, though.
The least interesting content is "here is how I burn tokens to build mediocre software".
The default workflow is 1) you create or derive documentation from the source code 2) PRD is derived from docs 3) tickets are derived from PRD 4) agents can work on tickets. All artifacts are files which live in the repo.
https://github.com/synlace/devtop
Edit: Still under active development.
However I would never use someone else's starter repo. The whole point is to augment my own perspectives and strengths. On top of that, the AI written pitch is extremely off-putting. If you can't even bother to take the time to write your own pitch, why would I believe for a second that your workflow has anything unique that goes beyond the lowest-common denominator AI capabilities.
There is danger in over-fitting the agent. Fast release cycles require frequent model switching, sometimes to another provider altogether.
In the end, it's hard to measure improvement. Most of the advice out there is vibes based, treat it as such.
You cannot write the files directly nor can you download them directly, but you can ask a chat to show you the content or only the mermaid diagram that exists in workspace.md (or however you want to name it), and mermaid can represent quite complex networks out of the box, their relations, what is blocking something else, and so on. Chats can then edit this file with search-and-replace instead of re-writing it all over again.
(I am a PM of 15 years trying to navigate all of this…)
I'm seriously worried that my and other companies in my network have product leadership that are wasting their mental bandwidth on AI fun stuff, needlessly writing their own random tools, etc - while the product scope is going in random directions with zero product vision. This is not about job protection or "stay in your lane".
For those that didn't read the article, one of the first things it says, in bold nonetheless, is "I didn’t design this system. I arrived at it"
The code that goes into PLC equipmentss and hardware is incredibly simple—but it's still a product. If you look under the hood of a lot of software, the code is a mess. Still, it's a product. And when Claude's code leaked, everyone laughed at the code hygiene—but a lot of people still used it.
What is a product, really?
What is a product? 106 points by rogix 4 days ago | 68 comments https://news.ycombinator.com/item?id=49202690
There is a lot of hand wringing in the software space about ai code being a mess - but from a product perspective code quality doesn’t matter. It’s the reliability of what people (or agents) use that matters.
Code quality used to further that goal but only because it was really hard to rewrite or fix code. If that barrier is removed does anyone really care what the code quality under the hood is?