AI coding assistants can help startups develop products, seed VCs believe

Today, there is nearly no developer on this planet who doesn’t use an AI co-pilot ultimately. However, using GitHub Copilot or Cursor.AI to ask technical questions and get debugging help could also be just the start. One day, AI coding may include agents that can write programs themselves based on natural language prompts. Such programs can even replace human engineers.

AI coding startups that can generate code based on natural language prompts include Replit and Bubble.

Ultimately, in accordance with some VCs, firms will employ fewer engineers and every of them will manage agents coding artificial intelligence. “It’s not a cake made in heaven. It’s in the near future, but not today,” VC Corinne Riley, partner at Greylock, said last week on stage at TechCrunch Disrupt.

She added that coding assistants are already widely accepted in coding interviews for potential employees at lots of Greylock’s portfolio firms.

However, he doesn’t believe that to be able to lower your expenses, really young firms should ever use artificial intelligence agents to exchange human engineers. At the seed stage, “you might be constructing the inspiration of the corporate, right? So when you’re making major engineering compromises at this stage, it’s probably not the fitting decision. These are decisions you can make in the longer term,” she said.

But money management can be why young startup engineers should enlist AI coding help as often and in addition to possible, Elizabeth Yin, co-founder and general partner of Hustle Fund, said on the VC stage.

“One of the main challenges in the early stages is that you don’t really know what problem you’re solving, what the ICP (ideal customer profile) is and what exactly they need. So you’ll end up throwing away a lot of work. So the faster you can work and the faster you can iterate, the better in terms of learning quickly,” Yin said.

He believes that early-stage startups ought to be open to any tools that allow founders to quickly assemble product samples to maneuver faster, even in the event that they should be fastidiously and thoughtfully rebuilt later. “I would actually be a fan of it if it meant you could learn a lot faster,” she said.

This is in contrast to the times before artificial intelligence, when each pilot needed to be coded by someone with the suitable skills. Today, an engineer can view a model, use AI debugging, and have a look.

VC Renata Quintini, early-stage co-founder of Renegade Partners, agrees.

“When it comes to finding product-market fit or testing it, that leverage needs to be leveraged, but I wouldn’t worry about seed-stage optimization,” she said on stage.

Interestingly, as startups founded in 2024 launched using AI development processes, we could witness the seeds of the primary future workforce of AI agents. And the primary people to recruit AI agents will likely be the programmers themselves. This thought is as ironic because it is prophetic.