AI Coding Agent Showdown — Ollama VS Gemini VS YouWare

This no-code platform has a surprisingly simple way of generating web apps.

They keep it simple. Really simple — every app is just a single HTML file.

No frameworks. No npm installs. No builds.

And the results are awesome. I’ll show you.

Check out the video on YouTube

You can sign up with my link and start creating AI apps on YouWare for free.

Course Update: New Style

I’ve updated the CSS for all lessons in my AI Engineer Roadmap course and Blog.

Feedback would be nice. Does anything not look right to you? Do you need a light mode?

(email me → alex@zazencodes.com)

Topics from this week’s video

• AI Coding Agent Platform Showdown

  • Goal: Build a travel planning map app using 3 platforms

  • Platforms tested:

    • Cline with Ollama and DeepSeek running locally

    • Gemini CLI (Gemini 2.5 Pro and Flash)

    • YouWare (sponsored, community-driven online platform)

• Application Requirements

  • Web app with:

    • Random location selection

    • 3-day itinerary

    • Map with pins on the left

    • Itinerary display on the right

• Cline (Ollama + DeepSeek R1 8B)

  • Setup:

    • VS Code + Cline extension

    • Configured for 128k context and 3000–4000s timeout

  • Observations:

    • DeepSeek model loaded despite exceeding local memory

    • Model created XML-tagged plans and basic index file

    • Attempted to use CLI tools correctly

    • Struggled with CSS and map rendering

    • No final working app output

• Gemini CLI

  • Setup:

    • Used Gemini 2.5 Pro (downgraded to Flash)

  • Observations:

    • Fast React project scaffolding

    • Struggled with execution and progress

    • Incomplete feature implementation

    • No map or working UI in final output

• YouWare Platform

  • Intro & Community:

    • Community-driven AI coding

    • Supports likes, comments, and sharing

    • Auto-complete while prompting

  • Key Features:

    • Auto-generated to-do lists (agentic pattern)

    • Built-in tools: web search, mapping, AI features

    • Used Leaflet.js and simple HTML

  • MCP Tools Demonstrated:

    • Huge Icons – added UI enhancements

    • AI App Tool – made itinerary generation random and unique

    • Google Maps Tool – enabled interactive maps (after fixing API key issues)

  • Final Output:

    • Functional travel itinerary app with:

      • Randomized trip suggestions

      • Icons and map markers

      • Custom themes (dark mode)

      • AI features embedded and monetized via user credits

  • Code Export:

    • Full project downloaded and versioned

    • Code includes prompt, manifest, helper scripts, and to-do tracking

    • Shared via custom YouWare link and GitHub

---

Literally every aspect of your day leaves a subtle imprint, like a potter’s hands pressing into clay.

Aaron Alexander