Overview
HealingMaps wanted to build an AI chatbot to answer questions about psychedelic-assisted therapies, treatment centers, and retreats – especially ketamine treatment.
Ketamine was granted “Breakthrough Therapy” status by the FDA in 2018 and quickly turned into a growing $5 billion industry. While ketamine therapy is in high demand, it’s still not easy to learn about online.
Psychedelic-assisted therapies have been proven to help people with mental health, addiction, and PTSD. But, ketamine, psilocybin, ibogaine, and MDMA and other psychedelics still come with a lot of stigma.
Some treatments are legal in some places and still illegal in others. This all makes it hard for people to find reliable sources of information about treatment.
That’s why HealingMaps exists. Their digital platform educates people about psychedelic-assisted therapies and connects them with legal treatments locally and around the world.
Problem
It takes deep expertise to educate people about psychedelic-assisted therapy. While scaling their company, HealingMaps wanted to give expert advice and deliver a personalized user experience.
To do this seamlessly for hundreds of thousands of monthly website visitors was a unique challenge.
They knew one new technology that could solve their problem – large language models (LLMs).
Since ChatGPT took the world by storm, LLMs have become more reliable and accurate. For example, Google just released the AIME medical diagnostic model that can diagnose patients via text as well as human primary care providers.
If HealingMaps could build a custom AI chatbot with augmented access to proprietary info about treatment centers they could educate people and connect them to legal treatment.
HealingMaps had three main goals to solve for:
- Create a novel chatbot using proprietary information about psychedelic therapies and make clinic recommendations based on user needs and location
- Build an MVP to show investors to continue to fundraise
- Engage AI development experts to build custom chatbot quickly without hiring internally
They found Codingscape and we quickly put a senior AI software engineer on the job to design and build their custom AI chatbot.
Solution
Custom AI chatbots are much more complex than they look. We needed to find a simple way to build a chatbot that scaled with HealingMaps as they grow. A big part of what made that possible was using Vellum AI behind the scenes.
Vellum is a development platform for building LLM apps with tools for prompt engineering, semantic search, version control, testing, and monitoring.
We used Vellum to give users access to a custom ChatGPT-experience with a geolocation API so the HealingMaps chatbot could answer all a user’s questions and then recommend a clinic close by.
You can use the chatbot on the HealingMaps website.
The whole application is hosted on AWS and if you want all the technical details you can see them below. Our senior AI software engineer (backed by our account manager) got the HealingMaps chatbot MVP up and running in 6 weeks. We delivered on time and within budget.
Now HealingMaps has a tool to educate their audience and connect people directly with reliable psychedelic-assisted therapy information and local treatment centers.
Engagement Details
Team Composition
- Account manager
- Senior software engineer
Duration
Project Stats
- Custom AI chatbot built with Vellum
- Delivered MVP in 6 weeks within budget
Technical Details
- Vellum AI creates prompts and workflows
- Application hosted on AWS
- React frontend website hosted on S3 behind a CloudFront cache
- Backend Express API hosted on an EC2 instance that uses PM2 to keep the process alive
- PM2 is behind an nginx proxy
- Geolocation API is a lambda function that talks to a MySQL Serverless V2 instance
Client Feedback
“As the founder of HealingMaps,
I had the privilege of collaborating with Codingscape to develop an innovative AI chatbot that has revolutionized the way we interact with our clients.
From the outset, Codingscape distinguished themselves with their exceptional expertise in AI and software development, coupled with a keen understanding of our unique needs in our healthcare sector.
I wholeheartedly recommend Codingscape to any business looking to leverage AI and dev solutions.
Their team's ability to translate complex requirements into seamless, user-friendly solutions is exceptional. They stand out for their technical excellence, commitment to client satisfaction, and ability to deliver on time and within budget.”
- Cory Jones, Founder & CEO HealingMaps
Sources
Technologies
AWS Cloud backend
- EC2
- S3
- RDS
- API Gateway
- Lambda
- CloudFront
- CloudWatch
- Terraform
LLM developer platform
- Vellum AI
Frontend
- React with Typescript
- Express with Typescript