I recently shipped Incident Analyzer, a small but powerful Next.js app that ingests security tickets and log files, then produces a NIST CSF-aligned incident summary, timeline, root cause, controls, and follow-up actions. The code is open source here: GitHub Repo.
It only took me 3 hours to go from nothing to live on Vercel. Along the way, I picked up a few lessons about using multiple AI tools together, and had a little “kitten chaos” moment that reminded me tech projects are never completely serious.
The Tech Stack
Here’s what I used:
- VS Code for editing
- Claude.ai (Claude Max) for untangling tricky TypeScript errors (Claude.ai is the one best friend that I do pay a pretty penny for, if you can upgrade, its worth it)
- ChatGPT for scaffolding and walkthroughs
- GitHub for version control and sharing
- Vercel for one-click deploy
- Canva for some simple UI polish
- OpenAI API (gpt-4o-mini) for the actual incident analysis – Don’t forget that you do need a credit card in hand to pay for credits if you never have purchased credits before.
Step 1: Starting From Zero
This was a computer that I normally don’t build apps on, so I didn’t even have a project folder set up much less VS Code or any of my normal tools. I did all my normal VS installs, including all the extensions I normally use, including Node.js 22.19.0, verified node, npm, and npx, then ran:
npx create-next-app@latest incident-analyzer --ts --eslint
cd incident-analyzer
npm i openai
I said yes to the src directory and App Router, which gave me the starter project.
Step 2: Writing the App
The app only needed two files:
src/app/page.tsx→ the UI (ticket input, file upload, analyze button, results panel, download CSV)src/app/api/analyze/route.ts→ the backend API that calls OpenAI and enforces the JSON schema
I pasted in the scaffolding from ChatGPT, but of course… errors showed up. That’s when the real lessons started.
Step 3: AI vs. AI Debugging
Sometimes one AI’s output won’t compile. I hit 6 errors in page.tsx and 2 in route.ts. Instead of burning hours debugging, I pasted the same files into Claude Max inside VS Code. Claude cleaned them up, aligned the types, and gave me a compiling v1.
Lesson: Don’t treat AI tools as “one and done.” Treat them like pair programmers. If one gets stuck, pit them against each other. That approach saved me time and sanity. (Sadly, I pay for 3 AI best friends, maybe 4, but don’t tell anyone!)
Step 4: Kitten Chaos
Halfway through, my progress suddenly stopped. Not because of TypeScript errors, but because my kitten decided to nap on my Loupedeck. She pressed the lock button, froze my PC, and left me staring at a dead screen until I figured it out.
AI didn’t solve that one. The moral of the story? Either disable the lock key, or accept that cats will always find the one button that halts your build. And will press it multiple times…
Step 5: Ship It
Once it ran locally, I pushed it to GitHub:
git init
git add .
git commit -m "first build"
git push origin main
Then I imported the repo into Vercel, added OPENAI_API_KEY as an environment variable, and deployed. That was it.
Live demo: incident-analyzer-seven.vercel.app
Reflections
- AI accelerates: Without AI, this would’ve been a full-day or longer project. With it, I had a working MVP in 3 hours.
- AI isn’t perfect: Expect compile errors. Use multiple models, cross-check outputs, and refine prompts.
- Humor helps: Sharing small hiccups (like the kitten mishap) makes the journey more real and a lot less sterile.
Try It Yourself
Clone the repo and run it locally:
git clone https://github.com/adaptoit-christi/incident-analyzer.git
cd incident-analyzer
npm install
npm run dev
Add your OpenAI API key in .env.local, then visit http://localhost:3000.
You’ll be running your own AI-powered incident analyzer in minutes.
In short: You don’t need weeks or a full dev team to build with AI. You just need curiosity, persistence, and maybe a backup plan for when your cat joins the project.
And if you don’t want to make your own and just want to play with what I built, here is the link to the tool: AdapToIT Incident Analysis Platform