Inspiration
During our in-person meet-up for this hackathon, Google had just released the Maps, Hotels & Flights plugins for their conversational AI, Bard. Upon learning that 3rd party extensions were coming soon for Bard, it was clear that our interests aligned in customizing the conversational AI experience for every day users. Conversational AI, especially models readily available to the public, has recently become massively impactful for just about anything you can imagine. A few of our team members had recently tried using AI to help plan their vacations in new locations, and were surprised at the accuracy & level of detail provided as the AI played Travel Agent for their benefit. The idea of having an AI interact with potential customers via direct chat and businesses via API started to take shape; for this hack, the business interaction is to book a reservation via Square Bookings API.
Proof of Concept
We kept our approach straightforward, as the goal was to test the options available to us via Google's AI tools and Square's API & Business platform, while building our toolbox and preparing for full 3rd party support down the line. From there, we landed on booking a reservation as a simple, low risk touch-point for customers & businesses. We created a mock AI modal so that we could leverage Vertex AI as our conversational AI, and hooked up our Square Bookings API directly to the AI, allowing the user to 1-click book an appoint in the chat window.
Challenges & Accomplishments
Square API documentation was great, and we utilized the testing live on the documentation pages. We tested and utilized the live demo features for several Google AI features. Google examples were easy to follow & allowed for quickly picking up different AI tools on the fly, allowing us to keep our options open for use cases. We workshopped ideas using the Image AI & the Document AI and were able to get started rapidly with the available resources. Ultimately, we decided to utilize the conversational AI for our proof of concept. Within the Vertex AI, the UI for adjusting the conversational AI parameters was clean, and interesting to learn about the effects on the Vertex model. The Square Developer platform was an area that we had the most problems with — consistent error message displaying on the appointments dashboard, and difficulty finding out the scope of all Square features (options hiding under ‘settings’ tab for each functionality).
Our biggest risk in the process was that the Square Appoints Plus free trial, which allows access to the Square Bookings API write access, is currently only limited time: this means from our hack start date to submission review date, we had to hot-swap credentials (or pay for write access to this API).
What's next for Google AI for Square Appointments
We will be following Bard updates closely, ready to hop on the 3rd party extension train. We aim to apply what we have learned of the Square & Google AI tools to build direct to consumer plugins, hone our skills & toolbox, and hit the marketplace running!
Built With
- amazon-web-services
- amplify
- bard
- go
- javascript
- lamda
- notion
- react
- square
- vertex
Log in or sign up for Devpost to join the conversation.