Sim AI on Mac: Local Installation, Real Test, and the Limits of the Open Source Version

▶

Sim AI, also known as SIM Studio, is one of the most interesting tools I have recently tested in the world of AI-powered automation. At first glance, it looks like something between ChatGPT and n8n: on one side, you have an AI chat where you can describe what you want to build; on the other, you have a visual environment where you can create workflows, blocks, agents, inputs, outputs, and automations.

The idea is very interesting: instead of using artificial intelligence only as a chatbot, Sim AI tries to turn it into a real workflow system. In practice, you are not limited to writing a prompt and receiving an answer. You can build reusable, editable processes that can be connected to other tools.

In this guide, we will see how to install Sim AI locally on Mac, which commands to use, which problems may appear during installation, and which limits I found in the self-hosted version compared to the cloud version.

What Is Sim AI and Why It Feels Like a Mix Between ChatGPT and n8n

Sim AI is an open source platform designed to create AI-based workflows. The simplest comparison is this: ChatGPT is a chat, n8n is a visual automation system, and Sim AI tries to combine these two ideas.

With Sim AI, you can create visual workflows, use AI models, manage inputs and outputs, work with blocks, and build small intelligent processes. For example, you can create a workflow that takes the text of a web page and generates an SEO analysis, or a system that turns a topic into a YouTube video outline, or even a process that classifies leads and prepares email drafts.

The important thing to understand right away is that Sim AI is not simply “ChatGPT installed locally”. It is more accurate to describe it as an AI workflow builder: an environment where you can build processes that use artificial intelligence models.

Is Sim AI Really Open Source?

Sim AI is available on GitHub and can be installed locally. This means you can download the project, run it on your Mac or on a server, and use the self-hosted part without completely depending on the cloud version. That said, during my test I found an important distinction: not all features available in the cloud version seem to be available or configurable in the same way in the local version.

The most obvious example is the Copilot chat, which is the feature where you can write something like “create a workflow to do this” and Sim tries to build the workflow for you. In the local version, this chat does not seem to automatically use your OpenAI API key saved in Secrets. To make it work, you need a specific key called COPILOT_API_KEY, generated from the Sim cloud version.

So, in my opinion, the most accurate way to describe it is this: Sim AI is open source as a platform and workflow builder, but some advanced features, such as the Copilot chat, still go through a service managed by Sim.

Requirements to Install Sim AI Locally on Mac

To install Sim AI locally on Mac, the first thing you need is Docker Desktop.

Download Docker Desktop here:

https://www.docker.com/products/docker-desktop/

Once installed, open Docker Desktop and make sure it is running. From the Terminal, you can check it with:

And:

If both commands return a version number, Docker is ready.

You also need Git, which is usually already available on Mac. You can check it with:

Download Sim AI from GitHub

Choose a folder where you want to install the project, for example Desktop or Documents:

Then clone the official repository:

Enter the project folder:

At this point, you are inside the directory where the docker-compose.prod.yml file is located. This is the file used to start Sim AI with Docker Compose.

Configure the .env File for Sim AI

Before starting Sim AI, it is a good idea to create a .env file inside the project folder.

You can create it with:

Inside the file, you can add a basic configuration like this:

To generate secure values, you can use:

Example:

Copy the generated value and use it for BETTER_AUTH_SECRET, ENCRYPTION_KEY, and INTERNAL_API_SECRET.

If you want to use OpenAI inside workflows, you can also add:

Be careful, though: this key is used for workflows and AI blocks. By itself, it is not enough to make the Copilot chat work in the local version.

Start Sim AI with Docker Compose

To start Sim AI, use:

The first time, Docker will download the necessary images and create the containers.

Once the startup process is completed, you can check the status with:

You should see containers similar to:

The migrations container may appear as Exited: this is normal, because it only prepares the database and then stops.

At this point, open your browser and go to:

If everything is correct, you will see the local Sim AI interface.

Redis Error in the Sim AI Copilot Chat

During my test, the local platform started correctly, but the chat remained stuck on “Thinking”. In the logs, this error appeared:

This means that the Copilot chat requires Redis, but the official Docker Compose file was not starting Redis in the configuration I was using.

To view the logs, you can use:

In my case, this was exactly the problem: Sim AI was running, the containers were active, but the chat could not respond because Redis was missing.

Add Redis to Sim AI with a Separate Docker Compose File

To solve the problem, I created a second Docker Compose file called:

Inside the sim folder, create the file:

Add this content:

Save the file.

At this point, stop the containers:

And restart using both files:

From now on, whenever you want to start Sim AI with Redis included, you should always use this command:

To view the logs:

Configure the Copilot Chat with COPILOT_API_KEY

Another important point is the COPILOT_API_KEY variable.

If you see a message like this in the logs:

it means that the Sim Copilot chat does not have the key it needs to work.

This key is not your OpenAI API key. It is a key generated by Sim AI in the cloud version. To get it, you need to log in here:

https://www.sim.ai/

Then go to the workspace settings, open the Chat keys section, and create a new API key for the chat.

Once generated, you must add it to your local .env file:

Then restart:

This is a very important point: the chat that generates workflows does not simply use your OPENAI_API_KEY. The Copilot chat goes through a service managed by Sim, and for this reason it requires a separate key.

Practical Test: Create an SEO Workflow with Sim AI

To test Sim AI, I created a simple but useful workflow: a mini SEO agent to analyze a web page.

The goal was to create a workflow that takes the description or text of a page as input and returns:

The prompt used in the chat was similar to this:

As test input, I used:

The test is useful because it clearly shows the purpose of Sim AI: not simply generating a text response, but turning a request into a reusable process.

The interesting part is not that Sim can write an SEO analysis. ChatGPT can do that too. The interesting part is that Sim tries to turn that logic into a workflow: input, processing, structured output.

Sim AI and n8n: Similar Tools or Complementary Tools?

Sim AI is often compared to n8n, but in my opinion they are not exactly the same thing.

n8n is very strong when you need to connect external tools: WhatsApp, Gmail, webhooks, Google Sheets, CRMs, APIs, databases, and different services. It is an excellent tool for creating operational automations between platforms.

Sim AI, on the other hand, seems more focused on AI reasoning inside a workflow. It becomes interesting when you want to create processes where artificial intelligence has to analyze, generate, transform, or classify information.

The difference can be summarized like this: n8n connects tools, while Sim AI tries to make AI reason inside a process.

For this reason, I do not see them as direct competitors. I see them more as complementary tools: n8n for connecting services, Sim AI for designing processes where AI plays a central role.

Privacy and API Keys: Pay Attention to Where Data Goes

During the test, I noticed an important aspect: the Copilot chat may ask you to enter an OpenAI key or information related to providers directly during the conversation. In my opinion, this opens a discussion about privacy and security.

If I am using a platform installed locally, I expect keys and data to remain as much as possible under my control. But if the chat part goes through a Sim cloud service, then we need to be aware that part of the experience is not completely local. I am not saying that Sim AI does anything wrong. The point is different: when we use self-hosted tools, we need to know exactly where credentials, prompts, client data, and workflow information are going.

For personal projects or tests, experimenting is fine. But for business data, real clients, sensitive API keys, or confidential information, this distinction becomes very important.

Is Sim AI Free? Open Source Software, but Real Costs

Another point that came up during the test is cost. The cloud version of Sim AI offers free credits, but even a simple test can consume quite a lot of them. In my case, with a fairly simple test, the usage was several hundred credits out of the 1,000 free credits available.

This means Sim AI should not be presented as “free” in an absolute sense.

The software may be open source, but there can still be real costs:

The most accurate phrase is: open source does not mean zero cost.

Sim AI Self-Hosted: What I Liked and What I Didn’t

What I liked the most is the idea. Sim AI tries to move artificial intelligence beyond the simple chat interface and into a reusable workflow logic. This is very interesting for those who work with SEO, content, web development, automations, and digital consulting.

I also like the fact that the project is available on GitHub and can be installed locally. For developers or people who want to experiment, this is a positive point. The less convincing part is the difference between expectation and reality in the self-hosted version. Installing Sim AI locally does not automatically mean having the entire cloud experience fully autonomous.

The Copilot chat requires a specific Sim key, Redis must be configured separately, and some features still seem connected to the cloud service. This does not make Sim AI useless, but it is something to understand before presenting it as “ChatGPT plus n8n, open source and free”.

How Sim AI Could Become Really Powerful

In my opinion, the most interesting thing, especially for developers and technical consultants, would be to take the open source part and adapt it to specific needs. For example, it would be interesting to connect the chat to providers directly chosen by the user, use local models, integrate Ollama, or build an infrastructure where API keys and data remain fully under control.

That would be the most interesting version for anyone who really wants a self-hosted AI system: not only a local interface, but also the model, provider, and orchestration logic under their own control. As it is now, Sim AI is promising and interesting, but I would not define it as a completely autonomous and free solution. It is more accurate to consider it an open source AI workflow builder with some cloud-managed features.

Useful Links to Install and Learn More About Sim AI

Official Sim AI website:

https://www.sim.ai/

Sim AI GitHub repository:

https://github.com/simstudioai/sim

Docker Desktop for Mac:

https://www.docker.com/products/docker-desktop/