How We Built Sense: The AI Tools Behind Our Indie App

A few years ago, building a polished consumer app as a solo developer meant either learning design, hiring freelancers, or shipping something that looked amateur. Today, the toolkit has completely changed.

I built Sense - a family organization app with AI - largely by myself. Not because I'm particularly talented, but because AI tools have leveled the playing field in ways that would have seemed impossible even two years ago.

This isn't a sponsored post. These are the actual tools I use daily. Some I built myself (because I needed them), others I discovered along the way. If you're an indie developer or thinking about building something, this might save you some time.

The Design Problem

Every app needs icons. Lots of them. App store icons, in-app icons, marketing images, blog thumbnails, social media assets. For Sense, I needed hundreds of visual assets across different contexts.

The traditional options weren't great:

  • Hire a designer: $50-200 per icon, weeks of back-and-forth, doesn't scale
  • Use stock icons: Generic, everyone else uses them too
  • Learn design: Months of practice to get decent results
  • Use Midjourney/DALL-E directly: Inconsistent styles, wrong formats, lots of iteration

I needed something faster. So I built it.

Pixelle

AI-powered icons and marketing assets for indie developers

Pixelle started as an internal tool. I was spending hours in Midjourney trying to get consistent icon sets, then manually resizing and formatting everything. It was tedious.

So I built a tool that generates app icons, in-app icons, and marketing images in the right formats and sizes automatically. You describe what you need, pick a style, and get assets ready to drop into Xcode or your marketing site.

The blog post thumbnails for our Cozi alternatives and OurHome alternatives articles? Generated with Pixelle in minutes. The app icons throughout Sense? Same thing.

I eventually released it publicly because other indie devs kept asking what I was using.

Try Pixelle

The Code Problem

Writing code is the easy part. Maintaining it, debugging edge cases, writing tests, refactoring - that's where solo development bogs down. AI coding assistants have changed this dramatically.

Claude Code

AI coding assistant that works in your terminal

Claude Code is my primary development environment now. It's not just autocomplete - it understands the entire codebase, can make changes across multiple files, runs tests, and commits code.

For Sense, I use it for everything from implementing new features to debugging production issues. The context window is large enough that it actually understands how different parts of the app connect.

This blog post? Written with Claude Code. The entire marketing site? Built and maintained with it. When something breaks at 11pm, I don't have to fully context-switch into the codebase - I can describe the problem and get a fix.

Try Claude Code

ChatGPT

General-purpose AI for research and ideation

ChatGPT is my research assistant. When I'm exploring a new feature area - like how other family apps handle chore rewards, or what parents actually struggle with - I start with ChatGPT conversations.

It's also useful for drafting marketing copy, brainstorming feature names, and getting quick answers to "how do other apps do X?" questions. Not always accurate, but good for initial exploration.

Try ChatGPT

Gemini

Google's AI for search-connected tasks

Gemini fills a different niche - it's connected to Google Search, so it's better for questions that need current information. Competitor research, checking what's trending in the App Store, understanding recent changes to APIs.

I also use it for tasks that benefit from Google ecosystem integration, like analyzing Google Trends data or understanding Search Console issues.

Try Gemini

What This Actually Looks Like

Here's a real example. Last week I needed to write a blog post comparing family chore apps. The old way:

  1. Research each app manually (2-3 hours)
  2. Write the post (2-3 hours)
  3. Create a thumbnail image - hire someone or spend hours in Figma (1-2 hours or $50+)
  4. Format everything for the web (1 hour)

The new way:

  1. Ask Claude Code to research the apps and draft the post structure (30 min of back-and-forth)
  2. Review and edit the draft (30 min)
  3. Generate a thumbnail in Pixelle (5 min)
  4. Have Claude Code create the HTML and add it to the site (10 min)

Total time went from 6-8 hours to about 75 minutes. And the quality is comparable - arguably better, because I can iterate faster.

The key insight: AI tools don't replace thinking. They replace the mechanical parts - the typing, the formatting, the repetitive research. You still need to know what good looks like, what your users need, and what direction to push the tools. But the grunt work? That's largely automated now.

What Still Requires Humans

AI tools are powerful, but they're not magic. Here's what still needs human judgment:

  • Product direction: AI can help you build faster, but it can't tell you what to build. Understanding user problems, prioritizing features, knowing when to ship - that's still on you.
  • Quality bar: AI output needs editing. The first draft is rarely good enough. You need taste to know what to keep, what to cut, and what to push back on.
  • User empathy: AI doesn't use your app. It doesn't feel the friction of a confusing flow or the satisfaction of a well-designed feature. That intuition comes from actually talking to users.
  • Debugging the weird stuff: AI is great at common problems. But the truly weird bugs - race conditions, device-specific issues, edge cases that only happen in production - still require human detective work.

The Indie Developer Advantage

Big companies are slow to adopt these tools. They have design systems, approval processes, legal reviews. A feature that takes me a day might take a team weeks just to get through the pipeline.

Solo developers and small teams can move faster right now. The AI tools favor people who can context-switch quickly, make decisions without meetings, and ship without permission.

This won't last forever. Eventually the big players will figure out how to use these tools at scale. But right now, there's a window where indie developers can punch above their weight.

Try the Tools

If you're building something, here's where I'd start:

  • For icons and marketing images: Pixelle - built specifically for app developers who need consistent, properly-formatted assets
  • For coding: Claude Code - the best AI coding assistant I've used, especially for larger codebases
  • For research: ChatGPT or Gemini depending on whether you need current information

And if you're a busy parent looking for a better way to manage family chaos, try Sense. It's the app all these tools helped me build.

Built for Busy Families

Sense is the family organization app I wanted but couldn't find. Try it free.