If AI can write most of the code now, where does the value of our software actually live?
That question took over a backlog refinement session with my team last week. It started with a quick comment about the Adobe clones built with AI and turned into a long discussion, with three positions that each had a good argument behind it:
- The value is still in the code, in algorithms tuned over years against real cases. Seeing an implementation saves a competitor years of discovery.
- The code is worth almost nothing now, and the value is in the people: a company's ability to keep the ones who hold the domain expertise and know how to make these tools work for them.
- The value lives in the workflow, in software that takes that expertise and applies it for the customer, so they get the result without configuring every step.
The more I thought about it, the more the three positions looked like the same domain knowledge kept in different places. It starts in people's heads, gets compiled into algorithms, and reaches the customer through the workflow.
AI made it much cheaper to turn that knowledge into code, and it did very little to make the knowledge itself easier to get or to replace. That's probably why the code argument and the people argument felt so opposed in the room, when they were describing the same asset at two different moments.
I'm still not sure which of these a team should protect first, and I suspect the answer depends a lot on the product.
Asking the PMs and my network here: in the era of AI, where do you think the value actually sits?