Intelligent AI Health Companion

The Challenge: 

A med-tech startup came to us with a big idea: developing an AI companion to empower all kinds of people to take control of their health through prevention and treatment advice. The AI companion needed to learn about the patient, provide relevant and factually accurate medical advice, and of course, maintain a high level of privacy. 

We began our journey by focusing on Alzheimer’s: helping both patients and their families get advice on how to manage the disease on the daily, as well as giving advice for those concerned about preventing Alzheimer's.

Our Solution:

We built a fully customised AI Chatbot API, designed not just to answer questions, but also engage in a conversation and ask questions to the user to fully understand their situation and needs and create a patient profile. 

Throughout the conversation, the AI Chatbot analyses responses and accesses a large knowledge database to provide precise and relevant answers. 

How did we do it?

• We first used our AI expertise to create a customised knowledge base consisting of hundreds of thousands of official medical journals and research papers: that ensures the chatbot would only provide medically accurate information.
  
  

• We then created the basis of our intelligent AI agent, giving it sophisticated instructions on how to gather the right amount of information from the user, how to use the knowledge base to provide recommendations, and how to handle sensitive information.
  
  

• Last but not least, we gave our AI agent a voice via an LLM (a “Large Language Model: basically, what gives the AI the ability to “speak”), instructing it on specific tones to adopt depending on the type of conversation and style of the user, making interactions feel smooth and empathetic.

Linking all these together, and our AI Health Companion was born! It can understand its users, ask them relevant questions to grasp their situation and provide them with relevant, factually accurate answers and advice. Essentially, an AI agent that doesn't, just write, but thinks!

Results:

With its intelligent, personalised and reliable advice, this AI Health Companion is a game-changer for personal health management: as of today, 95% of the recommendations from the chatbot match doctor’s recommendations.

Behind the Scenes - The Tech Note:

Getting a bit more into the technical details: 

Our API for this AI Agent is fully customised, meaning we created our own fragmented Agent Executor from scratch and linked it to a customised database. It’s also model agnostic, meaning it’s not tied to any particular LLM. 

Why? This setup gives both more control and flexibility by letting us give the Agent specific rules, tones, and the freedom to switch to another LLM on the fly; as well as ensuring the AI Agent sources from a filtered and secured set of information rather than the whole Internet. 

This setup is the one our experts recommend for Enterprise AI chatbot solutions, and the one we use to build customer service AI chatbots, e-commerce recommendations AI chatbots or educational AI chatbots; as it ensures full customisation and use of relevant-only information.