The ability to connect custom data sources (DataSource) to favorite apps is a must-have that greatly facilitates both Users' and Administrators’ work. Therefore, AI-UniBot Personal Assistant & Corporate Chatbot, our flagship product with over a hundred daily Users, was one of the first to undergo this upgrade.

Thus, our Clients got the opportunity to integrate any knowledge storage systems into AI-UniBot work without waiting for product updates. They can also adapt the chatbot to their unique IT landscapes, where data is often distributed across various platforms (from standard SharePoint or Jira to niche or proprietary solutions), even by the Client's IT Specialists.

DataSource regarding AI-UniBot refers to IT systems used for accumulating structured and unstructured knowledge. For example, in SharePoint, it can be documents and presentations. In Confluence, technical documentation. In Jira, tasks and comments. Previously, AI-UniBot supported only a limited list of sources (Google Disk, Azure Files, etc.). This caused a dependency on product updates to connect new systems.

Meanwhile, our Clients quite often needed to integrate AI-UniBot with specific or outdated systems – those that are not included in the standard support list. For example, the open HelpDesk platform OTRS or internal CRMs with their own API.

Previously, such requests required modifications to the AI-UniBot core. And this took weeks or months. Now, Customers can add new data sources promptly.

To connect a custom DataSource, a separate web service is developed that complies with the Lizard Soft API standard. This service is responsible for collecting data from the target system (e.g., OTRS) and converting it into a structured format. It can be deployed on any infrastructure – from the Client's servers to serverless architectures (AWS Lambda, Azure Functions) – and implemented even using low-code tools. During indexing, AI-UniBot contacts the service, receives «raw» data, extracts key entities from it (texts, metadata, links), and stores them in a semantic index. For example, if the source is OTRS, the bot indexes tickets, support responses, and statuses to answer questions that have already been resolved without Human intervention.

Among the use cases we have considered are:
There is a Pharmacy Network that uses its own inventory management system based on an old local server. Data on medications is stored in formats not supported by AI-UniBot. With the help of a custom DataSource, the pharmacy's IT Department created a microservice that transforms information from the server's CSV files into a JSON structure understandable to the chatbot. Thus, the Employees of the Pharmacy Network always have up-to-date info on the availability of medicines through queries like «How much ibuprofen is left in Kyiv? », and the bot automatically analyzes the indexed data.

A Consulting Firm integrated AI-UniBot with an internal knowledge base based on MediaWiki. Previously, Employees of this Company manually copied fragments of articles into ChatGPT for answers. Now, after connecting a custom DataSource via a web service in Python, the bot independently indexes all Wiki pages, including the history of changes. On the query «What changes were made to the Customer policy in 2024? » AI-UniBot forms an answer based on the latest revisions, referring to specific sections.

In short, with the support of custom DataSources, AI-UniBot Personal Assistant & Corporate Chatbot has «learned» how to work with any data sources – from legacy systems to open-source solutions. This eliminates dependency on our updates, reducing your adaptation costs. Therefore, instead of focusing on eliminating technical limitations, you can freely focus on your current business tasks.