How I Use AI to Get More Done Every Day

AI isn't just a buzzword anymore. It's a practical tool that can save you hours every day if you know where to plug it in. Here's how I use AI across my daily workflow as a developer.

Code Faster

AI-Assisted Coding

The most obvious use case. Instead of writing everything from scratch, let AI handle the repetitive parts.

// Prompt: "Create a debounce hook for React"
// AI generates this in seconds:

import { useEffect, useRef, useState } from 'react'

export function useDebounce<T>(value: T, delay: number): T {
  const [debouncedValue, setDebouncedValue] = useState(value)

  useEffect(() => {
    const timer = setTimeout(() => setDebouncedValue(value), delay)
    return () => clearTimeout(timer)
  }, [value, delay])

  return debouncedValue
}

Tools I use:

• Claude Code — For multi-file changes, refactoring, and working across a codebase
• GitHub Copilot — For inline completions while typing
• Claude.ai — For designing solutions before writing code

When to use AI for code:

• Boilerplate and repetitive patterns
• Converting between formats (JSON to TypeScript types, SQL to ORM)
• Writing tests for existing code
• Regex patterns (nobody should write regex by hand)
• Scaffolding new components or endpoints

When NOT to use AI for code:

• Security-critical logic — always review manually
• Business logic you don't fully understand yet
• When you're learning a new concept — write it yourself first

Debug Smarter

Instead of staring at an error for 30 minutes, paste the error and relevant code into AI.

Error: Cannot read properties of undefined (reading 'map')

Code:
const items = data.results
return items.map(item => item.name)

AI will immediately tell you data.results might be undefined and suggest:

const items = data?.results ?? []
return items.map(item => item.name)

Pro tip: Don't just paste the error. Include:

• The error message
• The relevant code block
• What you expected to happen
• What actually happened

The more context you give, the better the answer.

Write Better Commit Messages

Instead of "fix stuff", let AI analyze your diff:

# With Claude Code
claude commit

# It reads the diff, understands the changes, and writes:
# "Fix race condition in user session refresh by debouncing token requests"

This alone saves me mental energy every day. I used to spend time thinking about how to describe changes — now AI handles it.

Automate Code Reviews

Before pushing a PR, run your changes through AI:

Review this diff for:
- Bugs or edge cases
- Security issues
- Performance problems
- Code style inconsistencies

AI catches things like:

• Missing error handling
• SQL injection in raw queries
• N+1 query problems
• Unused imports or variables
• Inconsistent naming conventions

It won't replace a human reviewer, but it catches the easy stuff so your team can focus on architecture and logic.

Learn New Technologies Faster

When picking up a new framework or language, AI is like having a senior developer sitting next to you.

Instead of reading docs for 2 hours:

I'm building a REST API with Hono.js for the first time.
Show me:
1. Project setup
2. A basic CRUD route for "posts"
3. Middleware for auth
4. Error handling pattern

You get a working example in 30 seconds. Then you iterate from there.

The learning loop:

1. Ask AI for a working example
2. Read and understand it
3. Modify it for your use case
4. When something breaks, ask AI to explain why
5. Repeat

This is 10x faster than reading documentation linearly.

Write Documentation

Nobody likes writing docs. AI does it for you.

Write JSDoc comments for this function:

function calculateDiscount(price, tier, isAnnual) {
  const rates = { free: 0, pro: 0.15, enterprise: 0.25 }
  const discount = rates[tier] || 0
  const annualBonus = isAnnual ? 0.05 : 0
  return price * (1 - discount - annualBonus)
}

AI generates:

/**
 * Calculates the discounted price based on subscription tier and billing cycle.
 * @param price - The original price before discount
 * @param tier - The subscription tier ('free' | 'pro' | 'enterprise')
 * @param isAnnual - Whether the billing cycle is annual (adds 5% extra discount)
 * @returns The final price after applying tier and annual discounts
 */

Use this for:

• JSDoc / TSDoc comments
• README files
• API documentation
• Onboarding guides
• Architecture Decision Records (ADRs)

Daily Communication

Emails and Messages

AI helps me write clearer, more concise messages:

Make this more concise and professional:

"Hey, so I was looking at the ticket and I think we should
probably not do it the way it says because there might be
some issues with the database if we do it that way, maybe
we could try a different approach?"

Becomes:

"I reviewed the ticket and have concerns about the proposed
database approach. I'd like to suggest an alternative —
can we discuss in standup?"

Meeting Notes

Record your meeting, transcribe it with Whisper or similar, then ask AI to:

• Summarize key decisions
• Extract action items with owners
• Identify unresolved questions
• Format as shareable notes

Terminal and DevOps

Generate Commands

Can't remember the exact docker or kubectl command? Ask AI:

How do I see logs for a crashed pod in the "production" namespace?

kubectl logs -n production <pod-name> --previous

Explain Commands

Found a command in a script and have no idea what it does?

# What does this do?
find . -name "*.log" -mtime +30 -exec rm {} \;

AI explains: "Finds all .log files older than 30 days and deletes them."

Infrastructure as Code

AI is great at generating Terraform, Docker Compose, GitHub Actions, and other config files:

Create a GitHub Actions workflow that:
- Runs on push to main
- Installs pnpm dependencies
- Runs linting and tests
- Deploys to Vercel

You get a working .yml file in seconds.

Data and Analysis

Quick Data Transformations

Convert this CSV data to a JSON array:

name,age,city
Alice,28,NYC
Bob,34,London
Carol,25,Tokyo

SQL Queries

I have tables: users(id, name, email), orders(id, user_id, total, created_at)
Write a query to find the top 10 users by total spending in the last 30 days.

SELECT u.name, u.email, SUM(o.total) as total_spent
FROM users u
JOIN orders o ON u.id = o.user_id
WHERE o.created_at >= NOW() - INTERVAL '30 days'
GROUP BY u.id, u.name, u.email
ORDER BY total_spent DESC
LIMIT 10;

My Daily AI Stack

Tips for Getting Better Results

1. Be Specific

Bad:  "Make this code better"
Good: "Refactor this function to reduce cyclomatic complexity
       and extract the validation logic into a separate function"

2. Provide Context

Bad:  "Why is this failing?"
Good: "This Next.js API route returns 500 in production but works
       locally. Here's the code, the error log, and my env setup..."

3. Iterate

Don't expect perfection on the first try. Treat AI like a conversation:

1. Get the first draft
2. Point out what's wrong or missing
3. Refine until it's right

4. Verify Everything

AI is confident even when it's wrong. Always:

• Test generated code
• Verify facts and API references
• Review security-sensitive code manually
• Check that suggested packages actually exist

5. Use System Prompts

When using APIs, set a system prompt that defines the style and constraints:

You are a senior TypeScript developer.
- Use functional patterns, avoid classes
- Prefer explicit types over inference
- Use early returns
- No comments unless logic is non-obvious

What AI Won't Replace

AI is a multiplier, not a replacement. It won't:

• Make architectural decisions — You still need to understand trade-offs
• Understand your business domain — It doesn't know your users
• Replace code review — AI misses context that humans catch
• Debug production issues — It can help analyze, but you need to investigate
• Write perfect code — Always review, test, and iterate

Summary

The developers who will thrive are the ones who learn to work with AI effectively:

1. Code — Let AI handle boilerplate, tests, and repetitive patterns
2. Debug — Give AI the error + context and get answers in seconds
3. Learn — Use AI as a personal tutor for new tech
4. Write — Docs, commits, emails, and meeting notes
5. Automate — Config files, CI/CD, infrastructure
6. Review — Catch bugs and style issues before your team does

Start small. Pick one area from this list and try it for a week. Once you see the time savings, you'll naturally expand from there.

Ready for the next level? Learn how agentic AI handles entire workflows autonomously, or read 10 advanced tips to cut your Claude usage in half.