LiveAI
Strategy

AI writes the code. You — think about the user

Copilot writes functions, Claude generates modules, Cursor assembles apps. Meanwhile, developers spend days figuring out what to build. Here's how this shift changes the profession.

12 min read Intermediate Developers, team leads
✓ A practical workflow: how to think about users before writing code
сценарий

Sound familiar?

Wrote perfect code — but users can't figure out the interface

You're debating how to implement a feature, but not whether it's needed at all

AI generates code in minutes — but you spend days figuring out what to build

Lead writes a spec → developer implements → user is unhappy. Cycle repeats

Знакомая картина — разберём, что с этим делать ниже.

Three shifts

The developer of the future isn't the one who writes better code

When AI writes code in minutes, the skill of 'writing clean functions' loses market value. That doesn't mean it's not needed — but it's no longer a competitive advantage. Like knowing how to drive: important, but not what defines a profession.

Three shifts are happening. From features to value: instead of 'how to implement' — 'why does the user need this'. From specs to understanding: instead of 'build per spec' — 'understand what's behind it'. From coder to product engineer: the person who understands both users and systems becomes indispensable.

Example: the task is 'fix the folder hierarchy'. A coder sees a spec for dragging and dropping elements. A product engineer asks: who creates these folders? How many nesting levels do people actually use? What happens when there are 200 folders? This isn't a spec — it's a challenge to think through scenarios.

LiveAI — for product thinking

LiveAI is where developers keep notes about users, scenarios, and decisions. AI sees this context and helps find blind spots in UX.

User notes + AI context — ask 'where will the user get confused?' and get a specific answer

ADRs in notebooks — architectural decisions as context for AI. A new developer asks 'why this way?' — AI answers

Iterative critique — describe a feature, ask AI to critique the UX, refine, repeat

Workflow: Think through a feature before writing code

1

Describe the feature in a notebook

Write down what you know about the feature: why it's needed, who it's for, what the use cases are.

  • Task: 'Redesign navigation — users complain it's too complex'
  • Context: '80% of users are small businesses, 3-5 projects'
  • Current problem: '4+ levels of nesting, nobody uses it'
You're not writing a spec. You're capturing context — why this is needed and for whom. That's far more valuable than a specification.
2

Ask AI about the users

AI sees the feature context and can raise questions you haven't asked.

  • 'What skill level are the users who'll encounter this navigation?'
  • 'Where will a new user get confused?'
  • 'What navigation patterns work well for 3-5 projects?'
AI doesn't replace UX research. But in 5 minutes, you can uncover blind spots that you'd normally find a week after launch.
3

Map out use cases

Based on AI's responses, write down 3-5 key scenarios. This becomes the foundation for design.

  • Scenario 1: New user creates their first project
  • Scenario 2: Experienced user switches between projects
  • Scenario 3: User searches for a document in a large project
Scenarios bridge the gap between 'what to build' and 'why the user needs it'. Coders write functions. Product engineers write scenarios.
4

Ask AI to critique the UX

AI sees the scenarios and context — and can critique UX with specifics.

  • 'Find 3 weak spots in these scenarios'
  • 'What happens when there are 20 projects?'
  • 'What unusual scenarios am I missing?'
UX critique through AI is a fast way to find problems before code. More on this approach in the AI Critic guide.
5

Record an ADR

Write down the decision and reasoning. This is an ADR (Architecture Decision Record) — memory for AI and for the team.

  • Decision: 'Flat structure + search instead of deep hierarchy'
  • Reason: '80% of users don't use nesting beyond 2 levels'
  • Alternatives: 'Tree with drag-and-drop — rejected, complexity without benefit'
An ADR in a notebook is context for future decisions. When you return to the feature in a month — AI will remind you why you decided this way.

Workflow: UX critique through AI

1

Describe the current UX

Write down how the current interface works: what users see, what actions are available, where problems arise.

  • 'Users see a folder tree with 4 levels of nesting'
  • 'Dragging to move items, right-click for context menu'
  • 'Complaints: users lose documents, can't find the right folder'
Include real user complaints — AI will factor them into its critique.
2

Ask AI to find problems

AI sees the UX description and user complaints — and can suggest concrete improvements.

  • 'Why do users lose documents? Walk through it step by step'
  • 'Suggest 3 alternative approaches to navigation'
  • 'How do other products solve this problem?'
AI won't replace user testing. But it will help you generate hypotheses in minutes — instead of days of deliberation.

With context vs without

Without user notes, AI gives generic advice. With context — it critiques a specific scenario.

Feature: Navigation

AI sees scenarios, user complaints, ADRs. Critique is specific: 'Your approach breaks at 20 projects because...'

No context

'We recommend conducting UX research and A/B testing' — generic words you already know.

FAQ

Think about the user — AI will write the code

Describe a feature in a notebook, ask AI to critique the UX. 12 minutes — and you have scenarios and an ADR, not just a spec.

Related content

SolutionsDevelopers GuideAI Critic Guidesmart-notes-ai Comparisonvs Obsidian