š” Will Building with AI Eventually Require Fewer Developers?
So hereās something Iāve been thinking about lately: with the rise of AI coding assistants, boilerplate generation, and one-click scaffolding toolsā¦ are we heading toward a future where fewer developers are actually needed to build great products?
I donāt mean tomorrow ā but maybe 5, 10, 15 years out. If AI keeps improving (and honestly, itās already kind of insane), will we still need large engineering teams? Or will smaller teams ā even solo devs ā be able to ship like never before?
Iām not saying devs are going extinct (Iām a dev myself š ), but I do wonder:
Will early-stage startups hire fewer engineers?
Will AI handle more of the āgrunt workā while we focus on product logic and creative direction?
Could this shift what it even means to be a developer?
Would love to hear your thoughts, especially from those using AI tools day-to-day. Are you already doing more with less?
Do you think this is a good thing ā or something to be cautious about?
Letās talk about where this might be headed. š
Replies
AdBlur - Hide Ads, Focus on Browsing
I went from an agency of 18 developers to 3 developers and one intern: Cursor
We reduced our MVP shipping time from 35 days to 7 days flat. We ship small one page developer utility tools in a few hours.
AI is king for solo devs.
Now for design purposes we still hire designers. AI hasn't achieved the designer status yet.
We have subscribed to 15-20 AI tools in the last year. We spend 30x less than what we spent on junior developers, video editors, transcribers, proofreaders, accountants and managers.
Iāve seen this shift starting to play out especially in early-stage teams. Tools like GPT-4, no-code platforms, and automation builders have made it possible to move from idea to prototype without deep engineering resources.
In my own work, weāve been able to automate internal ops workflows and product experiments that wouldāve previously needed weeks of dev time now itās often a mix of prompt logic, connectors, and a review layer.
But once the surface-level build is done, maintaining and scaling still requires serious engineering judgment especially if you're touching sensitive data or edge-case-heavy flows.
So yes, I do think the definition of ādeveloperā is expanding. Smaller teams can do more early on but they still need technical depth once things get real.
Curious how others are balancing AI-accelerated speed with long-term scalability.