AI Agents using agno and Ollama

Agents example list:

1. Simple Ollama Agent The first basic agent that uses Deepseek R1 (8B parameter) model.

2. Finance Agent A simple finance agent using Yahoo Finance tool and Qwen 2.5 (1.5B parameter) model via Ollama.

3. Team of Agents The team of agents uses DuckDuckGo search and Yahoo finance to summarize analyst recommendation and share latest news for a company given its stock ticker symbol.

4. Image Agent The image agent reads an image and uses Llama 3.2 vision model to describe the image.

5. Knowledge Agent (WIP) The knowledge agent reads a PDF using PyPDF and uses ChromaDB (a vector database) to store embeddings (using Ollama embedder) and retrieve knowledge based on user queries.

---

Basic Architecture

---

Requirements

• uv
• ollama

---

Commands

1. Initialize the project

uv init

2. Create a Python virtual environment using uv

uv venv --python=3.13.2

3. Install agno (previously phi data) and ollama

uv add agno
uv add ollama

4. Pull Deepseek R1 (8B) and Qwen 2.5 (1.5B) models from Ollama registry

ollama pull deepseek-r1:8b

ollama pull qwen2.5:1.5b

5. Run the simple ollama-agent app

uv run ollama-agent.py

---

Finance agent using Yahoo Finance tool

• Run the finance-agent app

uv run finance-agent.py

---

Team of Agents

Finance agent works with Search agent to find the analyst recommendation for a company.

uv run agent-teams.py

---

Image Agent

Reads an image and uses Llama 3.2 vision model to describe the image.

Model used: Llama 3.2 (11B parameter)

---

Knowledge Agent (WIP)

Reads a PDF using PyPDF and uses ChromaDB (a vector database) to store embeddings (using Ollama embedder) and retrieve knowledge based on user queries.

Model used: Llama 3.2 (3B parameter)

---

License

This repository is licensed under the MIT License.