My Journey From Simple LLM Calls to Fully Agent App Relay on Documentation Only
From raw LLM calls to full agent systems with langGraph and chain, here’s my learning path and the patterns I collected in agentic-ai-engineering.
It All Started With Simple LLM Calls
A few months ago, I was experimenting with Ollama on my laptop. I was keeping things simple. I had a small script that sent a prompt and printed the response.
import ollamamy_prompt= str(input())response = ollama.chat( model="qwen2:7b", messages=[{ "role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": my_prompt} ]) print(response['message']['content'])
I felt excited. I ask questions, get responses. For basic experiments, it worked well enough.
The problems started when I tried building something more useful. I wanted an assistant that could use tools, remember previous actions, and handle tasks across multiple steps. Things became messy very quickly.
I had no memory between calls, no structured way to call tools, and every new feature meant I was copy-pasting and...
Copyright of this story solely belongs to hackernoon.com. To see the full text click HERE
