Last month, I was on a 3-hour flight to Thailand.
No internet. No hotspot. And of course — a last-minute bug that needed fixing.
Classic.
I had some logs and notes locally, but I needed more than that — context, code suggestions, and some AI-like brainstorming. Normally, I’d pop open ChatGPT/Claude/Grok and move fast. But this time, I was stuck staring at my screen, trying to recall obscure error patterns from memory.
It killed me.
Then, After few minutes debugging and head-bashing I remembered, long-back I had Ollama installed on my machine, even tried llama3 few times. And I had recently installed this thing called Goose, a local AI agent framework. I had setup recently along with a few local models like gemma-1.1b, deepseek-r1:8b, qwen2.5-coder, and mistral. I’d only played around with them briefly right after installing, never really diving deep.
So I opened the terminal. Ran Goose. Pointed it to the local LLM from Ollama.
goose --model ollama://localhost:11434/qwen2.5-coder:7b
Within a few seconds, I was chatting — offline — with an agent that could understand my task, analyze my logs, and help brainstorm ideas.
It felt like ChatGPT — but fully offline.That flight changed how I look at AI tools.
I realized Ollama gives you a local model, but Goose gives it a brain — tools, memory, structured reasoning, and task management. It wasn’t just responding to prompts. It was helping me debug, summarize files, and even suggest shell commands based on my logs.
What started as a travel headache became a real discovery.
I didn’t need:
- An API key
- A browser
- A network
I just needed a working laptop. Goose gave me modular prompts, task agents, and the ability to extend it with custom logic. Suddenly, I could:
- Load a local file and ask it questions
- Create a custom agent just for log analysis
- Add tools to fetch system info, run code snippets, and more
And it all ran on top of Ollama — fast, private, and offline.
That flight could’ve been wasted time. But greatful to Goose + Ollama, I got work done, stress-free, 30,000 feet in the air. If you’ve ever wished ChatGPT worked offline, or needed an AI assistant that runs with zero dependencies on the cloud, this is it.
Goose + Ollama = Your offline AI co-pilot.