I’m currently working on a new chatbot to help designers and creative professionals quickly reference color codes for html and css.
In addition to building and running custom chatbots for brands and businesses, I enjoy creating chatbots for my own interests.
Even if the functionality is super simple, at the very least, the exercise allows me to learn the latest platforms and run experiments that I would not necessarily do with client projects.
In the end, I’m able to provide my clients with a higher level of service, and I even have a new tool that saves me time or money (or both!).
This exercise was all about testing out the latest capabilities in the Azure hosting environment, and their bot services that run within it.
I thought I’d share the details of what I used to build it, and what I learned from the experience so far.
Chatbot Stack
Cloud Hosting Environment: Microsoft Azure (requires a subscription)
Bot Framework: Azure Bot Service
AI Component: QnA Maker
Getting started creating the bot is fairly straightforward, though there are some technical details including creating an App ID, and determining a development language (C# or Node.js).
The Azure Bot Service provides multiple bot templates as a starting point, which you can then customize to a large extent. The bot template choices are included below.
Azure Bot Service Templates
- Basic
- Form
- Proactive
- Language Understanding
- Question and Answer
For this project, I chose the Question and Answer template, since it provides all the structure I need. I’ve used this template before, and I did notice some improvements since then.
The best aspect of this was the AI component, which is the QnA Maker. This provides a built-in advantage of Microsoft’s own AI technologies so that your bot ‘learns’ over time, improving the effectiveness and accuracy of your bot.
I like to spend a lot of time testing and really customizing bot to have a real personality.
To me, it’s super important the bot is accurate and concise with its answers, and that it speaks the same language as the organization and industry it’s serving.
The time spent on these last steps of testing and adding character to the bot is invaluable, and is what I feel truly sets our bots apart from many others.
I plan to wrap this up and deploy it in the coming weeks. I’ll be sure to share details about the bot when it’s ready, in case anyone wants to interact with it.
In the meantime, if you’re interested in seeing what chatbots can do for your organization, send us your details here.
