Artificial intelligence monitoring startup ArthurAI Inc. is looking to make its presence felt in the generative AI industry with the launch of its first chatbot builder, Arthur Chat.

Announced today, it’s described as a turnkey, retrieval-augmented generation platform that companies can use to power more accurate and reliable chat-based applications that are fed on their own proprietary data. With the launch of Arthur Chat, companies can now create and deploy fully operational chat applications within just a few hours of conceiving them, the startup said.

The service has the ability to integrate with almost any large language model, making it extremely adaptable, the company said. In addition, it also provides strong protections via its integration with Arthur Shield, a tool that prevents proprietary data from being leaked outside of its customer’s servers, and safeguards against inappropriate content generation. Finally, it comes with built-in protection against so-called “hallucinations,” ensuring that chatbots won’t go astray and start making up answers on the spot.

When customers get started with Arthur Chat, they’ll be able to choose their preferred LLM as the foundation of their new chatbot. Then, they can use its RAG capabilities to link the chatbot to their internal datasets and knowledge bases to ensure it is always fed with the most up-to-date information it needs.

Their chatbots will be more flexible with Arthur Chat too, as it enables customers to switch quickly between different LLMs, based on the complexity or type of questions asked by users. In this way, companies can take advantage of the best functionality of all of the latest LLM technologies at once.

The ability to switch between different LLMs is important because the most popular foundational models are far from equal. Some of them have been designed with particular types of logic or modalities in mind, such as being able to come to a particular answer swiftly based on a particular set of parameters or produce lengthy document summarization based on swaths of data. Others are lightweight and designed to be less costly to train and deploy, while yet other models have been developed to preserve the privacy of the users for regulatory compliance.

Arthur Chat will be able to help companies choose the best LLM to fit their needs, using Arthur Bench. Launched in the summer, Arthur Bench is a tool that enables teams to quickly evaluate models, prompts and hyperparameters for generative AI text models so they can select the most appropriate one based on what they need it to do, and other considerations such as costs.

The company also runs the Generative Assessment Project, which ranks the strengths and weaknesses of LLM offerings on the market from industry leaders such as OpenAI, Anthropic and Meta Platforms Inc. and shares discoveries with the public about their metrics.

Arthur offered a number of practical applications where Arthur Chat can be used to create more reliable chatbot offerings. For example, a retailer could create a customized chatbot that’s able to answer questions about the latest line of fall clothing, delivering detailed information about the latest products it has in store.

In finance, a hedge fund might use Arthur Chat to assist its analysts, leveraging proprietary data to provide the latest insights on a specific portfolio or asset. It could also be used for customer support, delivering more dynamic and accurate responses and handling more complex queries by digging up the information it needs from product manuals and other proprietary data, the company said.

Another advantage of Arthur Chat is that it can be combined with the company’s generative AI model monitoring tools, which can help prevent phenomena such as “data drift” and ensure chatbots do not become less reliable over time.

Arthur Chief Executive Adam Wenchel said Arthur Chat brings together the company’s full suite of products and capabilities into a single turnkey package. “Arthur Chat significantly accelerates time to deployment while also ensuring that AI-driven answers are accurate, free from sensitive data, and aligned with a company’s values,” he said.

Image: ArthurAI