After three decades of building websites and web applications, you develop a sixth sense for client communication. Not because clients are difficult. Most are great. But because there's almost always a gap between what's said and what's meant.
AI can write your code. It can't read your client's email and understand what the project actually requires.
Here are the most common client emails, decoded.
"We just need a simple website"
What they said: Simple. Maybe 5 pages.
What they mean: They need a website, and they want it to not be complicated for them. The word "simple" describes their desired experience, not the technical scope.
Dig deeper and you'll find: a CMS so they can update content themselves, contact forms that route to different departments, an events calendar, integration with their email marketing platform, e-commerce for 50-200 products, and mobile-responsive design that "looks like Apple's website."
None of this is unreasonable. But "simple" it is not.
The lesson: Always ask what "simple" means to them. Their definition and yours are almost never the same.
"Can you just add one small feature?"
What they said: One small thing.
What they mean: They've been thinking about this feature for weeks and have a very specific vision. It involves user roles, conditional logic, email notifications, a dashboard, and "maybe an API so we can connect it to Salesforce later."
The word "just" in a feature request is a red flag. It means the requester has already decided it's easy. They haven't thought about the technical implications because that's not their job. It's yours.
The lesson: There are no small features. There are features with small interfaces and complex implementations. Your job is to uncover the implementation before quoting the price.
"We need it to look exactly like this other site"
What they said: Make it look like [competitor's site].
What they mean: They like the feeling of that site. The polish, the confidence it conveys. They probably don't actually want a pixel-for-pixel copy, and they almost certainly don't know that the site they're pointing at was built by a team of 12 over 8 months with a custom design system.
What they actually need is a site that gives their audience the same level of trust and professionalism. That's achievable. Cloning a $500K website on a $50K budget is not.
The lesson: Translate the reference into specific design principles (clean, modern, fast, trustworthy) rather than treating it as a spec.
"The last developer messed everything up"
What they said: The previous team did a terrible job.
What they mean: One of several things. Sometimes the previous developer really did bad work. Sometimes the client changed requirements repeatedly and the developer couldn't keep up. Sometimes the project was under-scoped and under-budgeted from the start. And sometimes the work was fine, but the client's needs evolved and nobody planned for that.
You won't know which until you look at the code yourself.
The lesson: Never take the "previous developer was terrible" narrative at face value. Audit the existing work, understand what happened, and form your own assessment. You're establishing the foundation for your relationship. Don't start it by inheriting someone else's blame.
"We're flexible on the timeline"
What they said: No rush.
What they mean: They have a launch date they haven't told you about yet. Maybe a board meeting, a trade show, or a funding milestone that depends on having a working product.
"Flexible" usually means "we don't want to pressure you, but we actually need this in 6 weeks."
The lesson: Ask directly: "What's driving the timeline? Is there an event or date we should be building toward?" You'll almost always get a real answer, and a real deadline.
Why This Matters for AI Development
Every email on this list represents a scope translation problem. The client knows what they want but describes it in their language. Which doesn't map directly to technical requirements.
This is the part AI can't do.
AI can generate code from specifications. It can't generate specifications from a client conversation. It can't hear "we need a simple website" and understand that means a CMS with role-based permissions and a product catalog. It can't read "one small feature" and recognize a six-week integration project.
This translation. From business need to technical scope to implementation plan. Is what experienced architects do. It's the reason projects with architect oversight deliver what clients actually need, not just what they literally asked for.
The 30-Year Pattern
After enough projects, you realize that most client communication challenges aren't about difficult clients. They're about the inherent gap between how business people think about products and how developers think about code.
Bridging that gap is the highest-value work in software development. It's not glamorous. It's not automatable. But it's the difference between a project that launches to applause and one that launches to a change order.
AI closes the gap between architecture and implementation. An experienced team closes the gap between what clients say and what they need. You need both.