Sherpa AI Server#
Sherpa AI Server is a web service application designed for training, using, and fine-tuning large language models (LLM) within the closed perimeter of corporations. Sherpa AI Server combines a document vector storage, offline model management, support for various artificial intelligence models – with different sizes, with quantization and without, their deployment on GPU or CPU, designing data processing chains, API access, a robotics platform, a web interface, and much more.
Sherpa AI Server is an Artificial Intelligence Center that contains functionality for:
· large language models (LLM) in a closed perimeter;
· web chat for company employees in the style of ChatGPT with history and dialogues;
· support for the Russian language;
· answers to questions based on proprietary documents;
· built-in document vector storage;
· API for any company applications, compatible with OpenAI;
· integration with Sherpa RPA - working with LLM from robot scripts;
· security, confidentiality, monitoring, auditing;
· multi-user and multi-threaded mode;
· choice from over 300 available language models;
· operation with CPU / GPU and the latest methods of quantization and batching of neural networks;
· the possibility of integration with any domestic RPA platform via API.
Sherpa AI Server has a wide range of capabilities and functionalities that can be applied as:
· a chatbot for customer support;
· a chatbot for internal user support;
· a chatbot for answering questions about corporate documents and knowledge bases;
· a document builder, contract generator, job postings, reports, analytical notes;
· a legal robot, sales robot, HR robot, interview robot, document management robot, etc.;
· extraction of structured and unstructured data from documents, including scanned ones, and entering them into information systems;
· generation of content plans, blog posts, articles, reviews, comments, press releases, digests, email letters, newsletters, presentations, etc.;
· generation of code, unit tests, macros, queries, comments, and documentation for code;
· semantic analysis and sentiment analysis of comments and customer feedback;
· text and voice BI (Business Intelligence) for corporate data;
· mass copying of document edits, tracking, and supporting changes in regulatory documentation.