The idea for SolvIT is to automate IT helpdesk with a chatbot. The primary features of the solution are: -Enable answering questions from internal knowledge bank -Address complex product “How to” questions that typically takes hours for IT analyst to figure out -Enable image processing to communicate effectively about an IT issue faced. -Collect Human feedback response. SolvIT can handle complex IT queries. It outperforms standard GPT-4 models by accurately synthesizing queries and delivering correct responses. this is enabled with RAG over a knowledge base. SolvIT excels at addressing both standard and custom application queries, leveraging GPT-4 Vision to interpret context from images/screenshots, ensuring accurate resolutions every time. We used Trulens to refine data inputs, benchmark performance, and evaluate tradeoffs. It's instrumental in troubleshooting issues, ensuring SolvIT development and improvement was measurable.
Doing an IT audit for an organization is a time consuming and very manual process. With this application, we have demonstrated the ability, using Gemini Pro, to automate a large portion of the tasks required in order to gather the required artifacts for the audit. Using the Gemini LLM and through integrations with external sources, we can determine the organizations posture with respect to asset patch levels for installed applications and the Windows operating systems. In addition, we have included the ability to track application license compliance through querying of internal systems and repositories. To address the use cases of specific queries that IT administrators have during an audit, we have implemented a chat bot to answer common questions as well.