Introduction

Onesait Platform makes use of Flowise, but before deciding on this tool, an analysis of existing tools was carried out to decide which was the best option for the Platform.

Therefore, during the third quarter of 2024, we analysed which tools to integrate for the creation of Intelligent Agents (IA), considering two options: Flowise and Langflow.

Candidates in brief

Flowise

This is an open source low-code tool for developers to create custom LLM orchestration flows and intelligent agents quickly and easily. It is built on top of NodeJS and has a multitude of tools to use in the flows.

In the following video the power of this tool is analysed and seen:

On GitHub you can download and install Flowise for local testing:

More information can be found on their website.

Langflow

This other web tool, also open source, is built in Python and offers a graphical interface to define LLM flows with LangChain, allowing to handle the concepts of ‘Chains’, ‘Agents’ and ‘Prompt Engineering’ in a very simple way.

This video gives you an idea of the power of the tool and shows you how to create a newsletter with Langflow quickly and with just a few clicks:

Langflow can be downloaded and installed on GitHub for local testing:

More information can be found on their website.

Analysis of the tools

Both tools are going to allow the creation of intelligent agents in a visual way, as a flow in a simple and fast way, through the use of drag & drop functionalities, but obviously there are differences between both of them.

The most significant of all is undoubtedly in terms of the general use of each of them. While both tools are very powerful, it can be said that Langflow is more oriented towards the creation of a chat bot, while Flowise allows a wider range of artificial intelligence applications to be covered.

In order to choose the final tool to be implemented, not only this important difference was taken into account, but also other points that were considered to be of special interest:

Interface

The user interface was intended to be simple and easy to use, which both tools fulfilled without problems.

Documentation

Without documentation you can go a long way, and again both have extensive documentation and a large community. From a Platform point of view, Flowise has a bit more detail, specifically in terms of configuration and deployment, which makes things easier.

For reference, the documentation for each tool is as follows:

• Flowise: Introduction | FlowiseAI

• Langflow: Welcome to Langflow | Langflow Documentation

Available components and extensions

At this point, Flowise has a larger number of specialised components. In addition, it offers the possibility of creating custom tools from NodeJS functions/modules, with the advantage that this implies.

Ease of integration with Onesait Platform

This is not a trivial aspect, because if it is approached from the point of view of maintaining a reduced and reusable ecosystem of technologies, in this case Flowise fits much better with the architecture of Onesait Platform.

The following aspects have been mainly assessed:

• Base Technology: Flowise is based on NodeJS, which will reduce the number of different technologies to use, as other components used in the platform, such as FlowEngine, are also based on NodeJS.

• Database: Both Flowise and Langflow require a relational database to store the workflow definition. In the case of Flowise, it has more diversity to configure, while LangFlow only allows the use of PostgreSQL. The choice of Flowise in this case will allow us to use our own configuration database (configDB). This simplifies not only the architecture, but also the deployments, as we will not have to manage different data stores, with the cost that this entails.