Conversational agents with data: the BIonA Search approach

Montevideo, August 6, 2025.

In recent years, conversational agents have evolved from simple text assistants to powerful interfaces for accessing information. A data-enabled conversational agent not only answers generic questions, but also connects to an organization’s real data sources to provide precise, contextualized, and actionable responses.

The great advantage of this approach is that it democratizes access to information: anyone, without technical or programming knowledge, can obtain reports, metrics, and analyses simply by asking a question in natural language, as if talking to a colleague. This removes barriers and speeds up decision-making.

In this article, we summarize the conversation with Santiago Atella, Director of the Intelligent Systems Division, about this approach and its challenges, and the aspects of these agents.

The challenges of connecting conversation and data

“Integrating a conversational agent with data is not trivial.” Among the most frequent challenges, he mentions:

• Heterogeneous sources: organizations have structured data (databases, spreadsheets) and unstructured data (PDF documents, manuals, internal websites).
• Semantic models: it's not enough to “search for keywords”; it's necessary to understand the meaning of queries and how they relate to the data structure.
• Security and permissions: information must be filtered according to profiles and roles, so each user accesses only what they are authorized to see.
• Visualization and action: responses must be more than just text; they should offer clear visualizations and options for deeper analysis.

BIonA Search: a smart conversational agent with real data

“In this context, BIonA Search —one of the modules of the BIonA Suite platform by Sofis Solutions— solves these challenges by offering an intelligent agent that understands and interacts with the organization’s data.”

Key features of BIonA Search:

1. Multiple data sources: integrates structured data (Excel, systems, databases) and unstructured data (webs, manuals, PDF documents).
2. Automated semantic model: interprets the query’s intent and maps the terms to the corresponding data structures.
3. Natural language: allows questions like “What were the sales in 2025?” and provides dashboards or reports without needing to write SQL.
4. Workspace-based security: ensures each user only sees the information they are authorized to access.
5. Visual and actionable results: generates dynamic dashboards and reports ready for analysis or decision-making.

How it transforms information management

“BIonA Search acts as a bridge between human language and the language of data,” says Ing. Atella. This means that:

• An executive can get a quarterly sales report in seconds, without relying on the IT team.
• An analyst can cross data from different sources and formats without manual consolidation processes.
• An operational area can access relevant manuals or procedures through a simple question.

“The transparency of results and the ability to combine real-time data make it possible not only to answer queries, but also to perform analysis and discover insights.”

A step towards organizational intelligence

“In a world where the volume and variety of data are growing exponentially, having a data-enabled conversational agent like BIonA Search is not just a convenience but a competitive advantage. It allows you to turn scattered information into accessible intelligence, optimizing processes, reducing response times, and strengthening decision-making.”

With BIonA Search, conversation becomes action and data turns into useful knowledge, directly accessible to any authorized user.