Just had an unexpected call yesterday with a potential client that left me thinking about our development philosophy at Uptech.

So here is what happened…

A lead reached out last week wanting to hire some of our dedicated developers for their project. This was actually the first time Ihtisham and I joined a client call together. We both hopped on the video call to make a good impression. The initial discussion went great, we talked about their requirements, our team’s expertise and everything was looking positive.

As you can see from our expressions in the picture, you can probably guess how the call eventually went

Then came this interesting exchange:

Client: “What IDE do your developers use?”

Me: “We work with VS Code.”

Client: “Why not Cursor? It would be much faster for development.”

Me: “We actually use GitHub Copilot and Claude AI to assist our developers, but…”

Client: “No, you should use Cursor. You just write a prompt and get an entire feature or module instantly. Your development speed would increase significantly!”

I tried explaining that while we embrace AI tools (all our devs have GitHub Copilot subscriptions and regularly use Claude AI for assistance), we are not willing to become completely dependent on automated code generation.

“For custom development projects, we need to deeply understand the code we deliver,” I explained. “While Cursor is great for web development, we have found that AI-generated code often requires significant reworking for optimal performance. We would rather write custom code with strategic AI assistance than depend entirely on automated generation.”

The client kept insisting: “But you would be much faster with Cursor. You’re making things too complicated.”

This got me thinking – are we really making things “too complicated” by not going all-in on AI-generated code? After reflecting, I’m even more confident in our approach.

Here’s why we maintain our balanced approach to AI coding tools:

• AI code often requires significant reworking for optimal performance and reliability
• Understanding the code at a deep level lets us debug and maintain it properly
• Custom solutions deserve custom thinking, not just prompt engineering
• The best software comes from developers who know when to write code and when to use AI

We ultimately had to agree to disagree. That potential client wanted developers who would rely primarily on AI coding tools, while we believe in using AI as one tool in our toolkit, not THE tool.

Sometimes saying no to a project is the right call when your development philosophy doesn’t align with a client’t expectations. We have built our reputation on quality custom solutions and that means being selective about how we work.

Have you ever had to politely decline work because your methods didn’t align with what the client wanted? I am curious how others handle this balance between AI tools and custom craftsmanship.