by Mike Williams
How I Built And Launched A Chatbot Over The Weekend
Take your idea to functional bot in hours, get real user feedback, and launch before the weekend is over! ?
Chatbots are not new. But they are a technology that most of us have probably read about, possibly used, and yet do not truly understand simply due to the lack of both everyday and prolonged exposure.
I wanted to change my understanding of chatbots and challenged myself to learn as much as I could over the weekend. I had determined it was time to really understand chatbots through mastering the fundamentals and building my own.
Like Most Projects and Companies I’ve Launched, I Began by Defining the Following:
- Timeline: Sunday was my deadline since this was a weekend project.
- Goals: Learn about the existing tools for chatbot developers that I could use to leverage technology and build on top of (I decided on Chatfuel). This would provide me with a fundamental understanding of chatbots.
- Success Outcome: Build a functional chatbot that did one thing well, iterate on real user feedback, and launch.
Unlike past projects, I didn’t have the exact idea in mind prior to making the decision to begin.
I had a window open on my laptop displaying a limited artist print I was looking at (Alec Monopoly’s “Pablo Judicial”). I simply typed “art chatbot” in Google to see if there was already a chatbot to purchase artwork.
To my surprise, there was not. I knew that my passion for art was the perfect match for my weekend desire to master chatbots.
I then removed the space in my search and had the name “Art Chatbot” in front of me. $9 and a few seconds later, and I also had the domain.
Since it was still early Saturday morning and the above had taken place in the matter of about 30 minutes, I had an hour or so to create the logo (using Squarespace logo creator).
Now that I had the logo, social cover, and also had the Facebook page live, I invited a few friends to like the page.
I got a few messages back asking what I was up to with the page and commenting on the name and logo. And so within minutes of my lean branding exercise, I already had some positive feedback!
I then told them what I was building and asked if they could test in the following hours over the course of my build.
This led to real user feedback as I was building so that I could create a more realistic conversational design and flow for my chatbot.
I realized that the process is fairly straightforward, and it was all about the conversational flow and user journey.
I Then Completed the Following Definition Exercise to Define What I Was Going to Build:
- Map out the conversational flow and user journey. In this case, it was for a user to discover artists, view artwork, and purchase artwork.
- Define the structure of the bot. In Chatfuel, this is done by “blocks,” which I created for each main state, sub states, destinations, queries, and answers.
- Identify and also limit the amount of content that I would display. Since I was building this in hours, it was more about the experience and validation, rather then how robust I thought it should be.
Actually building the bot in Chatfuel began at the second step with the blocks. It was mid-afternoon Saturday now and I was feeling a bit overwhelmed with the ambitious complexities I initially set out to build.
To overcome this, I removed some of the artists, the artwork to show, and then the artwork available so that I could focus on a better experience rather then trying to account for quantity.
Before Saturday was over, I had completed the initial version of Art Chatbot. I had friends testing the chatbot and providing real feedback, and all of the social pages and assets were prepped to launch before the end of the weekend.
With this initial version completed, I spent Sunday feeling a bit less time-constrained. I was able to focus more on prepping for a launch along with continual chatbot iteration.
I also began implementing basic AI functionality into the bot. I messaged a few art technology companies to ask what data was valuable to them (knowing I could easily collect this). This way, I could be on their radar and possibly build a chatbot that had some monetary value through data down the road.
I ended the day on Sunday by making a quick landing page using Instapage and optimizing it for a Product Hunt launch that I would do Monday.
To pre-launch the chatbot, I also setup Botanalytics (to help monitor, track, and analyze messages), posted in Facebook Groups to generate initial users, and then drafted my Product Hunt intro that I would post.
I also used a Google doc to create all the verbiage I would share on my social channels. Then I could simply copy the text, add the Product Hunt URL, and share across channels on Monday.
The result is that Art Chatbot hit 100 upvotes on Product Hunt and had also sent over 1,800 messages by the end of the day.
I not only learned about chatbots, but I am more confident in my understanding of the technology through a real use case. I also refined my ability and the process I used to build and launch a product rapidly.
Overall, I consider the weekend project a success — and more importantly, it was refreshing and fun!
I’ve also tried to “open source” everything about Art Chatbot’s build (I documented it on my social) and the chatbot roadmap. I will continue to keep everything publicly accessible (link below) going forward for anyone that is interested in being involved or contributing.
Art Chatbot Open Sourced ? bit.ly/artchatbotopensource
With an eagerness to learn and using the process above, it’s certainly possible for anyone to build their own chatbot in a weekend! ?
Be sure to tweet at me with any questions you have, feedback, and also let me know when you launch your own chatbot and I’ll check it out/retweet! You can also check out my personal chatbot I made after this weekend project here.
*A special thanks to Avery Andon and Art Life for the inspiration during the creation of Art Chatbot. Art Life is the official online gallery for both artists and artwork that are linked as URL’s in Art Chatbot.
Rather see a video format of this with more details? Watch this 12 minute video where I share more: