Introduction

Flowise has been integrated into Onesait Platform as an orchestrator of Large Linguistic Models (LLM) and custom Intelligent Agents (IA). With a low-code and open source approach, this tool allows developers to create their flows by dragging and dropping nodes of functionalities.

Open

Ejemplo de creación de nodos. Fuente: Flowiseai.com

At the language orchestration level, Flowise allows LLMs to connect with memory, data loaders, caching, moderation and more. It is fully compatible with LangChain, a framework that helps facilitate the integration of LLMs into applications, as well as with LlamaIndex, a GPT index for the data framework of LLM applications.

Regarding Agents and Assistants, it allows the creation of autonomous agents that can make use of tools to execute different tasks. These agents can use well-known services such as ChatGPT, or any other free service available (Ollama, etc.).

In addition, although it has a low code approach, it also has tools to extend and integrate the applications created through the use of APIs, an SDK and an integrated chat for the frontend that will allow the agent to be easily included in any web component.

Using LLM Flow

Create an instance

Flowise has been integrated so that users with ‘analyst’ or ‘administrator’ roles can create their own instances in this tool.

Each user will be able to create their own Flowise instance, which will be independent from the rest of the users. To do so, from the Control Panel, navigate to the ML & AI > LLM Flow Manager menu.

Open

The list of instances available to the user will then be displayed, which will initially appear empty:

To create the instance, click on the ‘+’ button at the top right of the screen:

Next, an identifier for the instance to be created must be entered.

Then click on the ‘Create’ button at the top right of the screen to finish generating the instance.

Once this has been done, you will return to the list of available instances, where you will be able to see the instance that has just been created together with information about it, such as its status (initially stopped) and other fields. You will also be able to see how the ‘+’ button has disappeared, since, as previously mentioned, only one instance per user is allowed.

Starting the instance and accessing Flowise

To start the instance, simply click on the ‘▶’ button in the options section.

Once done, the status will change to ‘Started’ and the CPU and memory usage will start to be displayed.

After starting the instance, to access Flowise, click on the ‘👁’ button, also located in the options section. This will open the Flowise flow panel:

To create a new chatflow, click on the ‘+ Add New’ button on the right hand side.

Once this is done, the flow creation canvas will be displayed: