The search and analytical functionality of UniBot Personal Assistant & Corporate Chatbot has been significantly boosted. Now, the new DeepSearch intelligent search algorithm searches for data, analyzes it, and summarizes everything studied. Thanks to it, at the User's request, the built-in Artificial Intelligence agents can make several queries to the knowledge base and even conduct QA sessions. This significant improvement allows UniBot to process complex queries even more efficiently and provide even more accurate and detailed answers.

Previously, UniBot had the Knowledge On-Demand mechanism implemented which proved itself well for relatively direct requests. For example, a User requests a comparison of two entities' properties, such as the capabilities of GPT 3.5 and GPT 4. UniBot is specifically examining the comparison materials. And if no comparison was found, the system replies that it does not have such information. This limitation became clear in cases where a more comprehensive analysis was required.

Now, thanks to the DeepSearch algorithm, the search AI agent, analyzing the first search results, can decide if the search has to be continued and in what way. For example, UniBot can independently change the search query, change the list of keywords it searches with, get the next page of results, or continue the search in a certain document that turned out to be very relevant.

Let's take the example of using it in the project management field. Let's say the Software Company Head of PMO wants to know what business processes the Project Manager is involved in. Before, if each business process was described by a separate document, UniBot could only find separate mentions. But it could not gather a comprehensive answer. Thanks to DeepSearch, UniBot can now process all relevant documents, extract information from each of them, and provide the User with a complete picture of all business processes the Project Manager participates in.

In this case, UniBot found that the Project Manager is involved in technical task development, risk management, interdepartmental coordination, and project quality control. By gathering all this data from various documents, UniBot provided the Head of PMO with a complete picture. Thus, the workload and effectiveness of the Project Manager were estimated much more accurately.

Another example is pharmaceutical research. Let's imagine: A Pharmaceutical Company Researcher asked UniBot to compare the effectiveness of a drug against hypertension over the past 14 years. Previously, UniBot could only provide individual studies conducted from 2010 to 2024, without their comprehensive analysis. Now, thanks to DeepSearch, UniBot can process all available studies, extract key data from them, and make comparisons. In this case, UniBot searched through all available scientific articles, clinical trials and reports, collected data on the effectiveness of the drug at different stages of the disease and among different age groups, and provided a comprehensive analysis. It included graphs, tables, and conclusions on performance dynamics.

Another use case is legal consultations. The Law Firm received a request from the Client to analyze the changes in the financial market legislation over the past five years. Previously, UniBot could only show separate laws or regulations. With DeepSearch, UniBot now analyzes all changes in legislation, extracts key points from each document, and provides a comprehensive overview of all the changes that have occurred during that period. This allowed Lawyers to quickly understand exactly how the legislation has changed and what consequences it might have for their Clients.

This is how it works: after receiving a request from the User, the search AI agent in UniBot v2 creates the first request in the semantic index. Next, using the DeepSearch algorithm, the Bot starts analyzing the results. If the first search does not provide a complete answer, the AI agent can change the search query, retrieve the next search page, or continue the search through specific documents. The answer will be provided to the User as soon as the AI is sure that it has collected the maximum information necessary to prepare such an answer. Therefore, the system provides deeper and more accurate data analysis. And Users receive more detailed and substantiated answers.

In short, the new DeepSearch algorithm in the UniBot v2 Personal Assistant & Corporate Chatbot significantly enlarges ways to improve information search and analysis. It allows you to get more complete and accurate answers to complex requests that require the processing of large volumes of data. So, thanks to UniBot, working with information gets as efficient, simplified, and convenient as possible.