← ContentsHome
Chapter 11

The Future of In-Situ AI

What's coming: multi-file refactoring, proactive suggestions, project-wide context, and runtime integration.

Where is this going?

What's Coming Soon (1-2 years)

1. Multi-file refactoring

Today: AI helps you change one function.

Tomorrow: "Refactor this module to use async/await throughout" → AI generates diffs for 15 files simultaneously.

2. Proactive suggestions

Today: You ask the AI for help.

Tomorrow: AI notices you're writing a test and suggests: "This test doesn't cover the null case. Want me to add it?"

3. Project-wide context

Today: AI sees your current file + a few related files.

Tomorrow: AI has indexed your entire codebase. It knows your architecture, patterns, and conventions without being told.

4. Runtime integration

Today: AI sees static code.

Tomorrow: AI connects to your debugger. It sees actual values, stack traces, and can suggest fixes based on runtime state.

5. Team context

Today: AI is personal.

Tomorrow: AI knows what your team is working on, recent decisions in PRs, and common pitfalls.


What Won't Change

Even as AI gets better, these principles remain:

1. Developers will still have limited focus

You can't increase the human cognitive budget. Tools must continue to respect flow.

2. Context will still matter more than model intelligence

A brilliant model without context is less useful than a decent model with great context.

3. Trust will still be earned incrementally

Developers will always review suggestions. The goal isn't to remove judgment—it's to reduce mechanical friction.


The North Star

The ideal AI coding tool is invisible.

You don't "go to" it. It's just there, like autocomplete, like syntax highlighting. It appears when you need it, helps quickly, and gets out of the way.

You stay in your editor. You stay in building mode. You stay in flow.

That's the north star.