Install Sim ai Locally with Ollama: Step-by-Step Tutorial

Unlock the next level of AI automation by building agentic workflows locally! In this comprehensive tutorial, you'll learn how to install Sim—an open-source AI agent workflow builder—on your own system, pair it with Ollama for local large language model (LLM) inference, and create powerful, flexible workflows that go far beyond simple chatbots.

Whether you're an AI enthusiast, developer, or business leader looking to extract true business value from agentic AI, this guide will walk you through every essential step, highlight important considerations, and offer actionable insights to help you get started confidently.

Why Move Beyond Chatbots? The Promise of Agentic AI Workflows

Traditional chatbots are just the beginning. Agentic AI enables intelligent workflows that:

Integrate multiple data sources
Interact with external APIs and other agents
Automate complex decision-making processes
Support parallel, sequential, and conditional operations

"If you want to extract business value out of agentic AI, you have to move beyond chatbots. That is where AI-powered agentic workflows come into play." – Fahd Mirza

Real-World Use Cases for Agentic Workflows

Automated customer support routing
Data-driven business process automation
Multi-agent collaboration for research or analytics
Custom LLM-powered toolchains

What Is Sim? Overview of the AI Agent Workflow Builder

Photo by cottonbro studio on Pexels

Sim is an open-source platform designed to let users visually build, manage, and run complex AI workflows. With Sim, you can:

Drag-and-drop modular blocks to define workflow logic
Integrate with over 60 pre-built tools and external services
Leverage advanced LLMs locally via Ollama
Apply enterprise-grade features with open-source flexibility (Apache 2.0 license)

Key Features of Sim

Visual workflow canvas for drag-and-drop design
Modular architecture with reusable blocks
Support for conditional logic, parallel operations, memory, and routing
Easy integration of local or cloud-hosted LLMs
Extensible with custom tools and functions

Prerequisites: What You Need Before You Begin

To install and run Sim locally with Ollama, ensure you have:

Ubuntu (or a compatible Linux system)
Conda (for virtual environments)
Docker (for containerized deployment)
Sufficient hardware (GPU recommended for best performance, CPU is possible for smaller models)

Tip: If you're new to Docker, check Fahd Mirza's easy-to-follow Docker installation video. GPU resources can be rented at a discount, as highlighted in the video.

Step-by-Step Guide: Installing Sim Locally with Ollama

1. Set Up Your Environment (0:00-2:30)

Create a Conda Virtual Environment
- Keeps dependencies isolated and organized.
```
conda create -n sim_env python=3.10
conda activate sim_env
```
Install Docker
- Required for running Sim and Ollama containers.
- Docker Installation Guide

2. Clone the Sim Repository (2:30-4:00)

Clone the Repo
- Get the latest open-source release from GitHub:
```
git clone https://github.com/<sim-repo-url>
cd sim
```
- Replace <sim-repo-url> with the actual repository URL as shown in the video.

3. Launch Sim and Ollama via Docker (4:00-6:30)

Start the Containers
- From the root of the repo, run the setup command:
```
docker compose up
```
- This will automatically download and configure Sim and Ollama.
Model Selection
- By default, Ollama downloads the Gemma model. To use a different model (e.g. GPT-OSS 20B):
```
docker compose run ollama pull gpt-oss-20b
```
- Modify the Docker configuration or select the model via Sim’s interface.

4. Access the Sim Web Interface (6:30-8:30)

Open your browser and go to http://localhost:3000
Sign up for your first session
Explore the visual workflow canvas:
- Left: Panel of workflow blocks (agents, routers, memory, tools)
- Top: Create or edit workflows

5. Build Your First Agentic Workflow (8:30-11:00)

Create a New Workflow
- Click 'New Workflow', give it a unique name, and start building.
Add and Configure Blocks
- Drag a 'Start' node and connect it to an 'Agent' block
- Select your preferred LLM (GPT-4, Gemma, or your custom model)
- Set system prompts, temperature, and other parameters
Enhance with Tools and Logic
- Add built-in tools or create custom tools (e.g., code execution, API calls)
- Use memory blocks, routers, and conditionals for advanced flows
Test Your Workflow
- Click 'Chat', input a message, and observe the model’s response, latency, and data output

Pro Tip: Start simple, then gradually add routers, memory, and multiple agents as you get comfortable.

Advanced Tips and Best Practices (11:00-End)

Expand Your Workflows

Integrate external APIs via custom tools
Use routers to branch logic and enable multi-agent collaboration
Add knowledge bases or document embeddings for context-aware responses

Optimize Model Performance

Use a GPU for large LLMs; consider renting if needed
Fine-tune models using resources like Transformers Lab (as mentioned in the video sponsor)
Monitor latency and resource usage for each workflow

Documentation and Community Resources

Explore Sim’s official documentation for advanced features
Join community forums or GitHub discussions for troubleshooting and ideas
Stay tuned for more advanced tutorials covering complex, real-world agentic workflows

Frequently Asked Questions (FAQ)

Q: Can I use Sim and Ollama without a GPU?

Yes, but performance will be limited for large models. Smaller models run on CPU.

Q: How do I add new models to Ollama?

Use the docker compose run ollama pull <model-name> command, then select the model in Sim’s interface.

Q: Are there enterprise features in Sim?

Yes, including pre-built integrations, conditional logic, and extensibility, all within an open-source Apache 2.0 license.

Conclusion: Your Next Steps in Agentic AI

By installing Sim locally with Ollama, you gain full control over your AI agent workflows—enabling secure, customizable, and powerful automation. Explore the drag-and-drop interface, experiment with different models, and iterate on your workflows to unlock the real business value of agentic AI.

Ready to build your own AI agents? Dive into Sim, start experimenting, and let us know your questions or experiences in the comments below!

Like this guide? Subscribe for more AI workflow tutorials and join the conversation!

Why Move Beyond Chatbots? The Promise of Agentic AI Workflows

Real-World Use Cases for Agentic Workflows