Agentic Engineering

How I Build Apps with an AI That Lives on My Server

Alexander Gomez · 2026

The Numbers Are In

90%
of code at Anthropic is written by AI

30%
of Google's new code is AI-generated

25%
of YC's Winter 2025 batch had codebases 95% AI-generated

These are not predictions. These are facts. From the last 12 months.

The CEO Mandates

• Shopify E-commerce, ~10% of US online retail
  CEO Tobi Lutke: "Prove why AI can't do it before requesting a new hire"
• Coinbase Largest US crypto exchange, NASDAQ
  Engineers who didn't onboard AI tools → Saturday meeting → fired
• Klarna Swedish fintech, "buy now pay later"
  5,000 → 3,000 employees. AI handles the work of 853 staff.
• Goldman Sachs Investment bank, founded 1869
  Piloting thousands of autonomous AI coders alongside 12,000 humans

The Tools Are Here

The question is: how deep do you go?

But Let's Be Honest

• METR study: experienced devs were 19% slower with AI — but believed they were faster
• Stack Overflow: only 29% trust AI accuracy (down from 40%)
• 66% spend MORE time fixing "almost-right" AI code

So which is it? Faster or slower?
It depends on how you use it.

Vibe Coding vs. Agentic Engineering

Two camps: Vibe coding (no review, prototypes) vs. Agentic engineering (AI does the work, you review and orchestrate)

I'm firmly in the second camp. And I'll show you why.

What is Claude Code?

Not a chatbot. Not an IDE plugin. An autonomous engineering agent.

• Lives in your terminal (or VS Code / JetBrains)
• Full codebase awareness — reads, searches, understands context
• Edits files, runs commands, commits to git, triggers CI/CD
• Plan mode, subagents, MCP servers, hooks

Think of it as: a junior dev that never sleeps, reads every file, follows instructions exactly… and never asks for a raise.

The Three Levels

Level 1 is talking to a tutor. Level 2 is pair programming.
Level 3 is having a developer on your team that you manage.

Where are most of you?

My Philosophy: Control + Plan Mode

• I like to be in control. I don't vibe code — I orchestrate.
• Plan mode: Claude reads the codebase, proposes a plan, I review, THEN it executes
• I review every change, approve every commit
• Result: consistent, less buggy code — the AI thinks before it acts

The Advanced Workflow

What Level 3 actually looks like in practice:

• Claude Code running on my VPS via SSH
• It commits changes → triggers GitHub Actions → auto-deploys
• Agent mode for deep research (parallel subagents)
• Memory files that persist across sessions (CLAUDE.md)
• I work from anywhere — phone, laptop, any terminal

My AI doesn't just suggest code. It ships code.

Act 3

Let Me Show You What I Built

Two projects. Both built with Claude Code.

Everything I'm about to show you was built by me and Claude Code together.
Not generated and forgotten — planned, reviewed, iterated, deployed.

EMX: What Is It?

"Turn your moments into cinematic soundtracks"

• Upload a photo → AI analyzes the scene
• Generates: lyrics, music, album art, music video
• Powered by: Gemini (vision + composition), Kie AI (music), Kling (video)

EMX: How It Works

• emx-api (Python / FastAPI)
  Vision analysis, lyric composition, image generation
• BFF (Node.js)
  SSE relay, orchestrates Kie AI for music & video
• emx-web
  Single-file SPA — yes, one HTML file!
• All behind Traefik reverse proxy, Docker Compose
• Real-time SSE events drive the progressive UI

EMX

emx.firebots.cloud

• Emotion Pad — built with Claude Code
• Upload a photo, watch face detection move the dot
• Walk through the pipeline as it happens
• Album cover, music, video — all generated live

Vibe Deployer

"Describe an app → AI builds it → deployed in 60 seconds"

• Chat interface at vibe.firebots.cloud
• Describe what you want in plain English
• Claude writes the entire app
• Auto-deployed to apps.firebots.cloud/{slug}
• Real-time progress via WebSocket

Vibe Deployer: How It Works

• Chat UI (nginx)
  Simple frontend, sends prompt via fetch
• n8n workflow
  Receives webhook → calls Claude API → writes HTML
• WS Relay (Node.js)
  WebSocket for real-time updates + file writer
• Apps server (nginx)
  Serves generated apps at unique URLs

Vibe Deployer

vibe.firebots.cloud

• Type a prompt — something fun
• Watch it generate in real-time
• Open the deployed app
• Show it actually works

A Day in My Terminal

• Multiple terminal tabs: Claude Code sessions for different projects
• Plan mode for anything non-trivial
• Agent mode for research: "find all the places this API is called"
• Git commits, PR creation, deployment — all from the conversation
• CLAUDE.md as persistent memory — Claude remembers your server

I don't just code with AI. I run my entire infrastructure with it.

The New Developer Desktop

• Tabs: Claude Code + Cursor + browser + Slack + monitoring
• Multiple agents running in parallel
• Background tasks: "research this while I work on that"
• Some devs run 3-4 Claude Code instances simultaneously

The developer of 2026 is a conductor, not a typist.

What I Learned Along the Way

1. Plan mode is non-negotiable
   "Measure twice, cut once" applies to AI too
2. You must understand the code
   You're the architect, not the AI
3. The AI makes mistakes
   But so do humans, and AI is faster at fixing them
4. It's not magic — it's a workflow
   The magic is in knowing when to use which level
5. It changed how I think
   I think in systems and architecture, less in syntax

The Timeline

My Challenge to You

You're all experienced developers. You know how to code.
The question isn't whether AI can help you.
The question is: at which level are you going to engage?

• Level 1: You're leaving 80% of the value on the table
• Level 2: You're getting there
• Level 3: Welcome to the future. It's already here.

Try Claude Code for one week. Use plan mode. Stay in control.
I guarantee you won't go back.

Resources

• Claude Code — claude.ai/code
• EMX — emx.firebots.cloud
• Vibe Deployer — vibe.firebots.cloud

Alexander Gomez

Questions?