SDE to AI Product Builder
Why this page exists
Section titled “Why this page exists”After 5 years of building distributed systems for other people, I’m learning to build AI products for myself. This page documents what that shift actually looks like — not the polished version, the real one.
If you’re an SDE thinking about building your own AI-powered products, this is the roadmap I wish I had when I started.
The Skills Gap
Section titled “The Skills Gap”Being a good SDE gets you maybe 40% of the way to shipping AI products. Here’s what’s missing — and what I’m actively working on:
1. Product Thinking
Section titled “1. Product Thinking”The gap: As SDEs, we build what the PM spec says. Building your own product means figuring out what to build, for whom, and why they’d pay for it.
What it actually means:
- Identifying real pain points (not imagined ones)
- Talking to potential users before writing code
- Scoping ruthlessly — shipping a small thing that works beats a big thing that doesn’t
- Understanding willingness to pay vs. “that’s cool”
How I’m learning: Building one product per week and seeing what sticks. Nothing teaches product sense like watching real people ignore your creation.
2. Design Sense
Section titled “2. Design Sense”The gap: SDEs can make things work. Making them feel good to use is a different skill entirely.
What it actually means:
- Layout, spacing, and typography that doesn’t look like a hackathon project
- UI patterns users already understand (don’t reinvent navigation)
- Knowing when “good enough” design is actually good enough
- Using AI tools (v0, Cursor) to bridge the design gap
How I’m learning: Studying products I use daily. Rebuilding UIs I admire. Using v0 to generate starting points and learning from what it produces.
3. AI Tooling
Section titled “3. AI Tooling”The gap: Knowing about AI tools is different from knowing how to build with them effectively.
What it actually means:
- Prompt engineering that actually produces reliable outputs
- Choosing the right model for the task (cost, speed, quality tradeoffs)
- Building AI features that handle edge cases gracefully
- Using AI coding assistants to 10x your own speed
How I’m learning: Using every major AI coding tool on real projects and documenting what works. See AI Coding Tools Compared for the living comparison.
4. Distribution
Section titled “4. Distribution”The gap: “If you build it, they will come” is the biggest lie in tech. SDEs rarely think about how people will find what they built.
What it actually means:
- Building in public (X/Twitter, blog posts, communities)
- SEO basics — making your product findable
- Launch strategy (Product Hunt, Hacker News, Reddit, Indie Hackers)
- Creating content that drives organic traffic to your product
How I’m learning: Documenting every build publicly on @gabrubuilds. Tracking what content drives traffic and what doesn’t.
Week-by-Week Learning Plan
Section titled “Week-by-Week Learning Plan”This isn’t theoretical. Each week maps to a real build and a real skill focus.
| Week | Build Focus | Skill Focus | Sharing |
|---|---|---|---|
| 1 | First AI tool (developer-focused) | AI coding assistants, rapid prototyping | Build log + tweet thread |
| 2 | User-facing product | Product scoping, talking to users | Build log + launch post |
| 3 | Design-heavy project | UI/UX, using v0 and design tools | Build log + design breakdown |
| 4 | Revenue experiment | Pricing, payments, value proposition | Build log + revenue report |
| 5 | Distribution play | SEO, content, launch channels | Build log + traffic analysis |
| 6 | Iteration week | Revisit best performer, add features | Build log + growth metrics |
| 7 | Integration product | API design, connecting services | Build log + technical deep-dive |
| 8 | Open source tool | Community building, documentation | Build log + GitHub launch |
Each week I’ll update this page with links to the actual build logs and what I learned.
Resources
Section titled “Resources”AI Coding Tools
Section titled “AI Coding Tools”These are the tools I use daily to build faster. Detailed comparison: AI Coding Tools Compared.
- Claude Code — Best for complex, multi-file projects and architecture decisions
- Cursor — Best AI-native code editor for everyday development
- v0 by Vercel — Best for generating UI components and full pages from prompts
- Bolt — Best for rapid full-stack prototyping in the browser
Product & Design
Section titled “Product & Design”- Indie Hackers — Community of builders sharing revenue and strategies
- Ship 30 for 30 — Writing/content framework (applicable to building in public)
- Refactoring UI — The best design resource for developers, period
- Mobbin — UI/UX patterns from real apps, great for design inspiration
Learning & Staying Current
Section titled “Learning & Staying Current”- Simon Willison’s blog — Best source for practical AI engineering insights
- Latent Space podcast — Deep dives into AI engineering
- Lenny’s Newsletter — Product thinking and growth
- Pieter Levels (@levelsio) — The blueprint for solo AI product building
Accounts to Follow on X
Section titled “Accounts to Follow on X”- @levelsio — Ships AI products solo, fully transparent with revenue
- @danshipper — Writes brilliantly about AI workflows and building
- @swyx — AI engineering, learning in public
- @mcaborern — Practical AI product development
- @gabrubuilds — That’s me. Follow along.
Follow the journey
Section titled “Follow the journey”This page gets updated every week as I learn and build. Bookmark it, or follow @gabrubuilds on X for real-time updates.