Chatbots have changed the world. Many times, it’s simply a more convenient and efficient way to interact with information. We felt that developing a chatbot for our clients would enhance the services we provide by enabling clients to find their personal health and program information quickly and clearly. Therefore, every client of Vivendi Health gets their very own chatbot through working with us.
How It Works
Over our time together, we will discuss many topics. Each of those conversations is recorded on your personal profile page. Your personal bot accesses your profile page and everything on it. The more time we spend together, the more valuable and rich this resource becomes.
We also export your lab and certain other diagnostic data and put that into a data table. This is a manual process for us but it delivers more reliable answers to queries, therefore we think it’s worth the extra effort on our part.
You can access your bot by saving a web app to your phone dashboard or from your web browser 24 hours a day.
Privacy
Your data is seen by you, your Vivendi Team, and no one else. The data is passed privately through Google Cloud Services into your OpenAi-based chatbot (Please see the bottom section if you’d like to review the full technical specs of your bot).
Data Sources
Not only does your bot contain your personal health information, it also contains our rich library of content that we have created (all in-house) over the last decade through humanOS, Better Brain Fitness, and more. We also provide access to ChatGPT through our bot so you can query the world, too, in case you’re looking for information outside our content corpus.
Sample Respones
Q: Show me my all my blood glucose readings
Here are your blood glucose readings from the data provided:
- November 13, 2023: 82 mg/dL (In Range)
- November 7, 2023: 75 mg/dL (In Range)
- August 21, 2023: 82 mg/dL (In Range)
- April 12, 2023: 89 mg/dL (In Range)
- January 23, 2023: 83 mg/dL (In Range)
- August 13, 2022: 83 mg/dL (In Range)
- March 19, 2022: 84 mg/dL (In Range)
- February 24, 2022: 89 mg/dL (In Range)
- December 9, 2021: 92 mg/dL (In Range)
- February 8, 2019: 83 mg/dL (In Range)
- April 13, 2018: 89 mg/dL (In Range)
- February 8, 2017: 93 mg/dL (In Range)
All your blood glucose readings are within the normal range.
Q: On my latest lab test, what values were out of range?
On your latest lab test dated 2023-11-13, the following values were out of range:
- Estradiol: 49.9 pg/mL (Out of Range-High; Normal Range: 7.6-42.6 pg/mL)
- AST (SGOT): 58 IU/L (Out of Range)
- ALT (SGPT): 45 IU/L (Out of Range)
Please note that the reference ranges for some biomarkers may vary between laboratories, and any concerns regarding these results should be discussed with a healthcare provider.
Q: What is my routine for Zone 2 training designed for me by my Vivendi team. Include what to consume before training.
Your routine for Zone 2 training should aim for 150 minutes per week, with your heart rate maintained between 111 to 125 BPM during activities like hiking, jogging, or cycling. It's recommended to do this work in a fasted state, which can enhance the metabolic response to exercise.
Before engaging in Zone 2 training, consume:
- Glucosamine: 1500 mg
- Collagen: 15 g
- Water and/or coffee
This will support joint health and hydration without breaking your fasted state. Remember, the goal of Zone 2 training is to promote mitochondrial biogenesis and improve cardiovascular health, doing so in a fasted state may further these benefits.
Q: Show me my average daily steps by year.
Here are your average daily steps by year for the available data:
- 2023: 11,045 steps (as of 2023-08-21)
- 2022: 12,079 steps
- 2021: 12,786 steps
- 2020: 13,639 steps
- 2019: 11,053 steps
- 2018: 10,283 steps
- 2017: 10,552 steps
- 2016: 10,912 steps
- 2015: 11,217 steps
- 2014: 11,867 steps
- 2013: 10,576 steps
- 2012: 11,189 steps
- 2011: 10,414 steps
These figures represent the average number of steps you took each day for each respective year. Please note that the 2023 data may not be complete for the entire year.
Making the Bot Ideal for You
Over time as we work together, we will make this bot ideal for you. Whatever you find most helpful and useful, we will craft your bot so it contains such information.
Technical Specs
The chatbot's backend development is handled in Python using the FastAPI framework, which ensures efficient and scalable backend requests and smooth communication with the frontend. It integrates APIs from OpenAI, Notion, and Google Sheets. The OpenAI API includes an Embeddings API for generating vector embeddings of user queries to search for the most similar data within the database and a Chat Completion API for generating responses to user queries using the GPT-4 model.
The Notion API is integrated to fetch and manage data from Notion databases, facilitating content and data retrieval, and the Google Sheets API is integrated with a Google Service Account to securely access data from private Google Sheets.
The frontend development is done in TypeScript using the React.js framework, which is effective for building dynamic and responsive user interfaces. WebSockets are used for real-time, bidirectional communication between the frontend and backend, which is essential for maintaining seamless user interactions and data flow.
The bot is hosted on the Azure platform, which provides comprehensive cloud services and ensures the high availability and scalability of the bots. It uses Kubernetes for containerization and orchestration, facilitating the management and scaling of containerized applications and improving deployment processes and infrastructure management.
The database used is MongoDB, dedicated to storing and efficiently searching for similar data using Atlas vector search. We opted for a dedicated plan on MongoDB Atlas, which meets production applications with sophisticated workload requirements and provides us with 10 GB of storage and 2GB of RAM, along with advanced configuration controls. MongoDB serves primarily as a repository for storing data and enabling advanced search capabilities using Atlas vector search. It operates as a document-oriented NoSQL database, providing flexibility in data modeling.
Finally, your chatbot is scheduled to be updated once daily through a cron job, but the Vivendi team can also manually trigger a sync when we update the information in your profile.