Although this application helps you to identify and discover your meal, its not really about food. The inspiration was to explore and learn how Appian, micro-services, and practical amplified intelligence can be combined to create a new and more compelling application and user experience.
What it does
Using food as an analogy you can shop, cook, go out, and just have your questions answered. Each case explores a different application of services and AI using Appian to put it together in a simple, compelling, and mobile experience. As just a sample of what can be done:
- Shop for food items by scanning bar code labels or typing in product names.
- Search for recipes by capturing information from an image of a recipe or just a picture of the food itself.
- Locate restaurants in your area using mapping and distance and duration measurements. Identify a restaurant and receive read reviews with sentiment.
- Ask food related questions like nutrition (carbs, fat, calories, etc) or you can ask for recipes that use an ingredient of your choice.
However these functions are just an analogy for all types of business to demonstrate that technologies can do new things like vision and understanding language.
How I built it
The application is assembled and presented through Appian which provided all of the user facing experiences (e.g. mobile user interfaces) as well as enabling of integration with a range of services including (Spoonacular, Google, Clarifai, and Intelexer). No additional components outside of Appian and the identified services were used.
Challenges I ran into
For my objectives, I did not hit many technical challenges. However. access to complete and useful sets of data was my largest challenge. I explored a range of services from Yelp, Zomato, OpenMenu, Wolfram Alpha, to name a few. Appian was able to easily integrate to these services but they did not have sufficient coverage to go as far as I wanted.
Additionally, I wanted to explore how custom component plugins could be used to create an even more compelling user experience. However, because of limitations on the ability to deploy such components to a cloud instance of Appian, I was not able to explore these technologies.
Accomplishments that I'm proud of
I started off with a very technical goal of exploring how to build an AI-centric application in Appian. But things came together in a surprising way and the application features integrated well together providing valuable information from a range of sources to consumers.
Additionally, this application an analogy for other business problems. Many of the concepts have explored also have application to a range
- Bar code scanning for identifying documents, identifying and tracking products, marketing, etc.
- Food data is like enterprise, commercial, governmental, and public domain data sources.
- Recipes are like scanned mail, email, chat, SMS text, and other documents.
- Ingredients are like enterprise products and service.
- Recipe collections are like product and service catalogs, customer relationship data, transactional systems, etc.
- Food is like products, places, faces, logos, demographics, moderated / NSFW content, etc.
- Mapping is like locating customers, vendors, landmarks, real estate, facilities management, asset tracking, etc.
- Street views are related to virtual presence, virtual tours, indoor mapping,
- Image labeling support classification of products, people, places, etc.
- Sentiment analysis for understanding and classifying customer tone in mail, email, chat, SMS text, etc.
- Conversational interface (chabot) is a more human natural interface - virtual assistants, self-service
What I learned
I am still new to the Appian community and find the low code promise of the software to be intriguing. I have used this project to deepen the skills I have in Appian and to push my self beyond what I think I knew.
Also, I have learned that intelligent services are much more capable than I previously understood. Many of these services appear to be ready now to add new and interesting features to a users experiences.
What's next for What's for Dinner
Additional features could be added to enrich and deepen the capabilities including finding improved access to menu information within restaurants, personalizing the application to maintain user favorites and preferences (e.g. dietary restrictions, favorite recipes, etc.).
More immediately I think there is value in sharing this prototype application internally within my company to evangelize the possibilities of how we can use Appian, micro-services, and amplified intelligence.