
Recent advancements in large language models (LLMs) have led to many novelty chatbot products, but very few have proven ready to meet the demanding requirements of the most sophisticated enterprises. After all, if Siri plays the wrong song in response to your request, you’re not likely to lose any customers. But if a chatbot provides incorrect, nonsensical and/or offensive responses to a customer, it can seriously damage ongoing relationships.
So, AI technology company Instabase recently announced the launch of AI Hub Chatbots, a product designed to overcome various challenges by enabling deployment in environments with highly stringent security requirements, supporting unstructured knowledge assets (including scanned documents, handwriting, etc.). This will provide reliable insights using an agent framework to solve complex multi-step tasks like financial analysis, and allowing users to verify every answer with references to the source knowledge.
In the announcement, Instabase noted that AI Hub Chatbots are already being deployed to solve important knowledge access scenarios, including in an air-gapped environment for a U.S. government agency. AI Hub Chatbots offer a number of advanced features, including:
- Anywhere Deployment: AI Hub chatbots can quickly be shared using various techniques ranging from links anyone can access for community users, in a secure SaaS environment, or even in air gapped on-prem deployments using Instabase’s self-hosted InstaLLM model.
- Broad Data Understanding: AI Hub can digitize and parse unstructured data. The solution will access digital PDFs, pictures of handwritten notices, complex tables and forms, or even websites to configure a chatbot in minutes.
- Confident References. AI Hub provides token-level, fine-grained references with confidence scores which allow users to verify responses with source documents.
- Complex Queries: AI Hub’s agent framework techniques allow chatbots to provide reliable insights backed by multi-step tasks like searching for information, extracting complex objects like tables and lists, using a calculator, critiquing its own answer, and recursively improving responses when they are incomplete.
“Our journey with AI Hub has shown us the immense value of knowledge access,” said Anant Bhardwaj, CEO and founder of Instabase. “Yet, we repeatedly see that business leads struggle to empower their teams to find insights in unstructured content. With AI Hub Chatbots, business leads can now instantly configure a chatbot on their knowledge base and share it securely with their teams or customers, providing immediate and reliable insights.”
Edited by
Alex Passett