My Journey Developing a Chatbot


I am building the first point of contact for anyone looking to learn about my history, experiences, interests, and hobbies

What I am doing

Explore and learn

A few weeks ago, I began building a chatbot. This chatbot’s purpose was not to manage a call center or provide QA support for an organization, but rather it was built to act as a virtual representation of myself. My goal was to have this chatbot (CalvinBot) act as an online point of contact for everything I believe and know. This would not be a complex chatbot along the lines of Alexa or the Google Assistant, but instead, it would be a simple QA bot with mastery on who I am, what I do, and my hobbies and interests.

It is a means of automating the common questions that people may have when they first meet me.

Why I am doing it

Value, automation, and ethics

I would love to say there is a compelling business value or challenge that CalvinBot is solving, but the main reason I built it was to develop my own experience and knowledge of chatbots.

Trending Chatbot Articles:

1. Knowledge graphs and Chatbots — An analytical approach.

2. Blender Vs Rasa open source chatbots

3. Picture my voice

4. Chat bots — A Conversational AI

As chatbots become increasingly prevalent in our lives I wanted to develop a first-hand experience on its value and limitations. What are the common questions people have when they want to learn more about me? How would they perceive a chatbot as their source of information? Do people need to meet me to know me?

Some questions I am sure people have for me

Chatbots offer a wide range of benefits from their 24/7 availability to their standardization, but I wanted to know if these benefits were desirable to consumers. Does changing the method in which people can reach out to me improve their overall experience? Would someone looking to connect with me prefer to develop an initial perspective of who I am using a chatbot?

Ultimately, building a chatbot allows me to make business applications ‘real’ by proving what it can and cannot do. AI, as a broad technology concept, poses the danger of being overinflated and overhyped. It can provide tangible value, but only for those willing to test it and learn about what it’s integration could mean for their organization.

What I am learning so far

  1. Building a chatbot is far easier than expected! Platforms like Dialogflow remove the need to code, and there are great tutorials on Youtube
  2. Training your chatbot is 90% of the work. I built my chatbot in half a day, but have spent weeks getting people to ‘break’ the chatbot so that I can train it on new interactions.
  3. ‘Crowdsourcing’ the act of training Calvinbot has allowed me to train it on intents I would have never originally considered (i.e. what colour is his eyes, what is his biggest regret).
  4. Chatbots should not be viewed as a replacement for human interaction. Every interaction that Calvinbot has had with people was a focused conversation with a clear agenda and outcome. However, this does not reflect the real conversations people have. Calvinbot does not handle ambiguity well and human-to-human interaction would provide the emotional nuance that Calvinbot cannot.

What is next?

Train, train, train!

I am continuing to train Calvinbot on new interactions with a better understanding of how people communicate. The end goal is to use Calvinbot to augment and support my networking ability, but not replace human-human interaction.

In the future, I will update this article with the key learnings I have noticed in the application of Calvinbot to human-led interactions.

Try it out!

If you want to see the current state of Calvinbot click the link below!


Disclaimer: This article was prepared or accomplished by Calvin Tennakoon in his personal capacity. The opinions expressed in this article are the author’s own and do not reflect the view of any employer

Don’t forget to give us your 👏 !

My Journey Developing a Chatbot was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

Source link

Related posts

Could OpenAI’s ‘too dangerous to release’ language model be used to mimic you online? Yes, says this chap: I built a bot to prove it


ONC Announces Winners of Certified Health IT Data Challenge


Human-AI Learning Performance in Multi-Armed Bandits. (arXiv:1812.09376v1 [cs.AI])


This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy


COVID-19 (Coronavirus) is a new illness that is having a major effect on all businesses globally LIVE COVID-19 STATISTICS FOR World