Category Archives: AI

AI

Gemini 3: Your New AI Coding Assistant

Every developer has that moment where they stare at the screen and wish for a magic wand.
Something that can unscramble a legacy codebase, sketch a UI without endless Figma tabs, or summarize a 300-page API doc that reads like… and create some good tests out of nothing.

Google just dropped something dangerously close.

Gemini 3 isn’t another “slightly better benchmark” release. It’s a real step forward—especially for people who build things for a living.

Here’s where it gets interesting:

Continue reading →
Standard
AI, webdev

8 Top Tips to Actually Use Cursor (Without Setting Your Wallet on Fire)

TL;DR

Open a new agent/chat for each new request or task, instead of continuing everything in one long conversation.

Keeping unrelated work in a single long chat makes the agent less efficient, increases context noise, and also becomes more expensive because the full conversation history keeps getting processed.

Best practice:

  • One request / task = one new agent
  • Keep chats focused on a single issue or feature
  • Start a fresh chat when switching topics
  • If you can –> Combine Cursor with Ollama + local models

If you’ve been coding anytime in the past year, you’ve probably heard the buzz about Cursor — the AI-powered IDE that promises to write your code, clean your code, and maybe even refactor your soul.

It’s built on top of VS Code, so it feels instantly familiar.
But the moment you hit that shiny AI shortcut, you realize: this thing is smarter than your codebase and hungrier than your wallet.

After a few months of using Cursor — and after accidentally vaporizing a scandalous number of API tokens — I’ve learned how to stay productive and solvent.


And yes, the TL;DR is that you can still combine Cursor with Ollama + local models to get many of these benefits for free.
Here are my 8 hard-earned tips to make Cursor your loyal sidekick within the limits of your budget.

The #1 tip: Control context scope aggressively – This is the biggest win

Cursor auto-includes files, diffs, and history—this explodes token usage.

Do this:

  • Manually select only the exact files/functions needed
  • Avoid “entire repo” context unless absolutely required
  • Use @file and @selection instead of implicit context
  • Clear chat or start a new thread when switching tasks

Why it matters:
Token cost scales with every line in context, not just your prompt.

Below are a bit more tips:

Continue reading →
Standard
AI

Transform Your Coding with Gemini CLI: A Local AI Assistant

Every developer has a moment mid-flow when they break concentration to look up a flag, debug an error, or Google that one awk trick they always forget. It’s death by context switching — and the browser is the grim reaper.

That’s where Gemini CLI comes in.

It’s not just another AI chatbot ported into a terminal.
It’s an embedded, context-aware development assistant that lives alongside your code, speaks your language, and remembers what you’ve worked on — locally.
No browser tabs, no copy-paste gymnastics, no handing your project to the cloud gods.

When choosing an AI coding assistant, developers have several strong options to consider.
Claude Code offers sophisticated reasoning and natural language understanding, excelling at complex problem-solving and architectural decisions through its command-line interface.
OpenAI Codex, which powers GitHub Copilot, integrates seamlessly into popular IDEs and has been widely adopted for its reliable code completion and suggestion capabilities.

Google’s Gemini stands out with its multimodal capabilities and strong performance across various coding tasks, while offering a particularly appealing advantage for developers just getting started: it’s available for free.

This makes Gemini an excellent entry point for newcomers who want to explore AI-assisted development without any initial investment, allowing them to experiment and learn before committing to paid tools as their needs grow.

Let’s unpack how Gemini CLI changes the game for developers, how to use it effectively, and where it still falls short.

Continue reading →
Standard