Day 161: I built my first chatbot in just 48 hours, here’s how it looked…

“If you’re not embarrassed by the first version of your product, you’ve launched too late” – Reid Hoffman

True to my word, today marked the day that Tinkerdash’s first chatbot was released. And true to Reid Hoffman, it was basic. Very basic in fact.

I’ll first explain what the chatbot does and why, and conclude with my thoughts on using Chatfuel to build it. So, without further ado, here is how the chatbot looked.

This slideshow requires JavaScript.

In a nutshell, the first version of the chatbot was a “lite”, semi-automated version of the full Tinkerdash service, which was provided to the bloggers during their review. It would serve to help users with their meal planning and shopping by providing meal plans catering for up to four people, eating three or five meals per week, based on a specific dietary need.

It would not allow users to customise what’s in their meal plan or assist with the ordering of their ingredients, though both of these things were planned for the future. And it would cater only for veganism, due to the time required to create the meal plans, with a view to expanding to the other special diets.

We opted to narrow the focus as described above because, based on what we learnt during the blogger review, this appeared to be a customer problem that we were able to solve with the resources at hand and thus get the next version of Tinkerdash to market quickly. As user feedback came in, the chatbot could be improved accordingly.

Here’s the general flow and a list of the screens:

  1. A welcome and introductory message, as well as an opt-in message to receive notifications of new meal plans and features (not shown).
  2. A basic navigation from where the user could select to “See meal plans”, “Share” and provide “Send feedback”. This was used as the central reference point.
  3. A menu to select a dietary preference. The only option in the first version was veganism, however, the user could register their interest in another special diet.
  4. A couple of questions to collect user preferences regarding the number of meals and servings they desired. This informed the meal plan and shopping list that they would see and help to segment the users in future.
  5. The ability to see the actual meal plan and shopping list for that week, opening as a PDF within the Messenger app. The PDFs were stored on Google Drive.
  6. A “Buy ingredients” button, leading to a message stating that this functionality is not available but is being worked on. This was to gauge interest for this feature.
  7. A “Share” option, allowing users to easily share Tinkerdash with their Facebook friends.
  8. A “Send feedback” option, allowing users to provide written feedback about their experience (or send smiley poo emojis, thanks everyone!).

I acknowledged that there was still a heck of a lot to do to make this into a kick-ass service, however, for something that was knocked up in a couple of days, I was pleased with the result. Which brings me nicely on to my thoughts on using Chatfuel to build it.

Chatfuel had been nothing short of amazing. I’ve not a bad word to say about it. It was functionally rich, easy to use even for a non-techy like myself and free (forever apparently!). And for more advanced chatbots, there was a whole host of integrations to connect APIs and other platforms.

I’d highly recommend it for people thinking about building a chatbot. Even if you’re not, you might just have some fun, instead of looking at cats or whatever.

On to the chatbot version deux…!

Comments are closed.

Up ↑

%d bloggers like this: