AI can build faster than ever. But are we losing the user?
by•
With tools like Claude Code, we’re entering a world where products can be built end-to-end inside AI workflows.
But there’s a problem I’ve been thinking about: As building gets faster, user research risks getting left behind.
At Velozity, we’ve been working on solving this by launching an MCP server that brings user research directly into these environments.
Now inside your coding workflow, you can:
Query personas
Access real interview transcripts
Ground decisions in the actual user context
Instead of switching tools or relying on assumptions.
We believe that if AI is going to build products, it needs to understand users just as deeply.
Would love to hear how others are thinking about this shift.
35 views
Replies
This hits something I’ve been struggling with lately. I can build quickly now but I often pause and wonder if I’m solving the right problem. Bringing real user context directly into the workflow feels like something I personally need. It helps me slow down in the right way while still moving fast.
Spot on, AI speeding up builds is thrilling, but skipping user research leads to shiny flops that nobody wants. From my side, I've seen founders prototype fast with AI but pivot 3x because assumptions trumped real user signals. Your approach flips that: query transcripts mid-prompt for evidence-based tweaks.
not sure this is really about access. most teams already have some kind of research lying around
just doesn’t get used that much in day to day decisions. wonder if putting it closer to the workflow actually changes that