I've been building web applications for over 15 years. Python backends, PHP systems, countless CRUD interfaces. Lately, I've been asking myself a question that keeps me up at night:
Is everything I know about building software interfaces about to become obsolete?
The rise of generative AI has me wondering: if users can just tell a computer what they want instead of navigating through my carefully crafted forms and menus, what's the point of traditional web development anymore?
After thinking deeply about this (and having some great conversations), I want to share my perspective. Spoiler: it's more nuanced than "AI will replace everything" or "nothing will change."
The Mechanical Nature of Current Computing
Here's what struck me recently: the way we interact with computers today is fundamentally mechanical.
Think about it. When a user wants to generate a sales report, they don't just ask for it. They:
- Navigate to the Reports section
- Click through dropdown menus
- Select date ranges
- Choose specific metrics
- Filter by region
- Click "Generate"
- Export to the right format
We've built systems that force users to adapt to the machine's way of thinking. They need to learn our mental models, navigate our menu structures, click buttons in our predetermined sequences.
And I've spent years perfecting this mechanical dance.
What AI Actually Changes
AI promises to flip this paradigm: instead of users adapting to machines, machines adapt to users.
The old way:
Navigate → Select → Filter → Configure → Execute
The new way:
"Show me Q3 sales performance in APAC compared to last year"
It's moving from explicit manipulation to intent expression.
But here's the key insight I've arrived at: This doesn't make traditional software development obsolete. It transforms it.
The CRUD Doesn't Disappear - It Gets a New Front Door
When a user asks that natural language question about sales data, something still needs to:
- Validate they have permission
- Query the right databases
- Aggregate the data correctly
- Handle edge cases
- Ensure data integrity
- Format the response
- Log the action for auditing
All the "mechanical" backend work I've been doing? Still absolutely essential.
The AI interface is just a new way to invoke these operations. Someone still needs to build the underlying system that can fulfill the request.
Beyond Text: The Full Picture of Interface Evolution
While everyone's focused on ChatGPT-style text interfaces, the transformation is actually much broader. Here's what's really changing:
1. Multimodal Interfaces
Not just text - combinations of voice, vision, and gestures:
- Point at something on screen + say "make this blue"
- Show a photo + ask "find similar products"
- Draw a rough sketch + describe what you want
What this means for developers: We're not just building forms anymore. We're building systems that understand intent across multiple input types.
2. Spatial Computing (AR/VR)
Apple Vision Pro and similar devices are pushing interfaces into 3D space:
- Gesture-based interaction
- Virtual screens in physical space
- Manipulating 3D objects naturally
The shift: From thinking in 2D screens to 3D spatial interfaces.
3. Voice as Primary Interface
Moving beyond "Hey Siri, set a timer":
- Natural conversations with context
- Voice + visual working together
- Speaking to accomplish complex tasks
The reality: Voice for input, screen for verification. Hybrid approaches win.
4. Ambient/Invisible Computing
Computing fading into the background:
- Systems that anticipate needs
- Context-aware automation
- Less "tell the computer," more "computer observes and assists"
The challenge: Building systems that are helpful without being creepy.
5. Agent-Based Interactions
The real game-changer:
- AI agents operating applications on your behalf
- Less "use the interface" more "delegate to assistant"
- Agents talking to other agents
Example: You say "Plan a weekend trip to Portland under $500" and an agent books flights, hotel, restaurants - no forms, no clicking.
What Actually Stays
After thinking through all these changes, here's what I believe will persist:
Structured Interfaces for Complex Work
- Developers will still use IDEs, not just chat with AI
- Accountants will still want spreadsheets for financial modeling
- Designers will still want visual tools
- Power users prefer precision over conversational ambiguity
The Underlying Architecture
- Databases, APIs, business logic, state management
- Security, validation, error handling
- Performance optimization
- Data modeling
Someone still has to build the house - AI just adds a different front door.
Hybrid Approaches
The future isn't "conversational OR structured" - it's both:
- Natural language for discovery and simple tasks
- Structured interfaces for precision and complex workflows
- Visual feedback of system state
- Keyboard shortcuts for power users
Think about it: calculators didn't eliminate spreadsheets, spreadsheets didn't eliminate databases. Each serves different needs.
The Middle Ground: Mechanical Becoming Organic
Here's my synthesis: The mechanical parts don't disappear - they get hidden behind more natural interfaces.
Instead of users learning the mechanical operation, the AI learns to orchestrate the mechanical parts. But someone still needs to:
- Design those mechanical parts
- Make them reliable
- Handle edge cases
- Ensure security
- Maintain data integrity
Your job as a developer shifts from "build forms" to "architect AI-augmented systems."
What This Means for Experienced Developers
Before:
Build the form → Build the validation → Build the API → Build the database queries
After:
Build intent recognition → Route to appropriate functions → Execute operations → Return results in natural format
Plus:
- Design APIs that AI agents can reliably call
- Build guardrails for AI-generated actions
- Create hybrid UIs (conversation + structure)
- Handle ambiguous inputs gracefully
- Ensure security when AI is the intermediary
The Skills That Matter More Now
Your experience with:
- Edge case handling
- Data integrity
- Security considerations
- Performance optimization
- System architecture
- Understanding business logic
My Honest Prediction: 5 Years Out
Here's what I think we'll actually see:
Many more conversational interfaces for simple, common tasks
- "Add this transaction to my expenses"
- "Schedule a meeting with the team next week"
- "Show me my top customers this quarter"
Traditional UIs for complex, professional work (with AI assistance built in)
- IDEs with AI pair programming
- Design tools with AI suggestions
- Spreadsheets with AI-powered analysis
Lots of hybrid approaches - chat alongside structured forms
- Start with conversation to clarify intent
- Switch to structured form for precise details
- AI fills in defaults based on context
The Bottom Line
The mechanical thinking isn't wasted - it's the foundation.
AI adds a natural language layer on top, but all that foundational work still matters enormously. Understanding data integrity, handling edge cases, building secure systems, optimizing performance - these skills don't disappear. They become the invisible infrastructure that makes natural interfaces possible.
We're not being replaced - we're being augmented.
The CRUD operations don't go away. They just get a smarter, more natural front door. And someone who understands both the backend complexity AND the new interface paradigms?
That person is incredibly valuable.
Questions for You
I'm curious about your perspective:
- How are you thinking about these shifts?
- What concerns you most about AI's impact on development?
- What excites you?
- Are you already building AI-augmented interfaces?
Let's discuss in the comments. After 15 years, I'm still learning - and I'd love to hear your thoughts.
Top comments (0)