The 2nd Best AI Chatbot on Earth

The 2nd Best AI Chatbot on Earth

A week ago one of our partners sent us a link to an AI chatbot that was actually very, very, very good. First of all, it was super conservative on bandwidth when embedded on a site. It had no superflous HTTP requests, and no "downloading half the internet to initialise the chatbot" stuff. The thing even had a couple of features we didn't have, and I have to admit, these were some very interesting features.

Oops, we're #2 in the world 😢

I can handle almost everything. I can handle that others are better than us on marketing. I can handle others being better at sales. I can even handle others being better than us at SEO and SMO - But there is one thing I simply cannot handle.

I cannot handle others having a better product than us!

And this was one of those cases where somebody had somehow magically managed to create a better product than us. However, that was yesterday - Today we're (of course) back on top again 😂

Back to #1 again 🥳

I spent a couple of days coding, taking some well deserved vacation from selling and marketing, and reversed engineered all features these guys had. Then I implemented all of these features into our AI chatbot. Below are the additional features and changes I had to do to make sure we ended up back on top again.

1. The chatbot that knows "everything"

This one was simple. The other AI chatbot would somehow magically know "everything", allowing it to answer questions it was not trained on.

This pussled me since we're so used to making sure our AI chatbots always stays on topic in regards to its RAG database to avoid AI hallucinations, that turning off strict context didn't even occur to me.

Initially I thought they would spice the chatbot's model with base data from some sort of database of industry knowledge they had, when it finally dawned upon me; "AI RAG 'leaking'."

RAG and VSS "leaking" to the rescue

Allowing the AI chatbot to "leak" is a dual edged sword, because it allows OpenAI to answer questions your RAG database doesn't know the answer to. At the same time, it allows the AI chatbot to use GPT4-turbo and its "base knowledge" to answer questions. For 99% of our customers this is probably not a good thing, since most clients wants to use the AI chatbot exclusively with their own data - But for some clients, especially clients with small amounts of data, it can be an advantage.

We're running a test on our own AI chatbot currently to allow it to answer all sorts of "unrelated questions" to see how it performs. In the screenshot below you can see how it answers a question we have zero training data related to.

Our AI chatbot answering questions about Bill Gates

Notice how our AI chatbot still does its best to guide the user back into the topic again, by asking relevant follow up questions, which brings me to the next point.

2. Fluent conversations

In lack of a better word, I refer to this as "fluent conversations". It implies that once the AI chatbot has answered your question, it will suggest follow up questions - Something you can see above with our Bill Gates screenshot. You can also try it below.

Every time you ask the chatbot a question, it will end its response with 2 to 4 follow up questions, allowing you to simply click these questions to continue asking more questions about the subject at hand. This is a super cool feature, especially if you're on a phone, where manually writing questions is difficult - And/or you simply want to investigate products and services the AI chatbot is configured to answer questions about.

Thank you dear competitor, that was an amazing cool feature suggestion - I'll be looking out for more features from you in the future 😂

Psst, please send me examples of amazing chatbots you find on email - Such that I can blatantly steal all their best ideas, and implement into our stuff 😊

3. Price

This one was a bit rough. Not only did the other guys have a kick ass AI chatbot, with the same quality as us, in addition to a couple of really cool features we did not have - But they were selling their product for $100 per month.

This was easily fixed by creating a "basic onboarding plan" that we charge $49 per month for. This plan is of course useless for 98% of our customers, since they need integrations, customisations, theme changes, etc - But for some website out there that simply wants a basic AI chatbot, with lead generation, our $49 per month offer, including OpenAI API tokens, should permanently eliminate the debate.

So now we're twice as good (again!) - In addition to that we're selling our AI chatbots for half the price 😊

Wrapping up

Considering we've got "a bajillion" additional features the other AI chatbot does not have, such as ...

  • Shopify integration
  • Lead generation
  • Questionnaires to collect data
  • Multiple themes and the ability to create your own theme
  • Dashboard with statistics, data exports, and KPI charts
  • Administration dashboard for managing your AI chatbots
  • WhatsApp, Messenger, SMS integrations
  • Etc, etc, etc ...

... I would claim the debate about who's got the best AI chatbot in the world should be permanently resolved (again!)

But thank you for some amazingly great ideas dear competitor - Let me know when you create another release, such that I can blatantly steal new features from you in the future 😁

And no, I will not tell you who these guys were - Simply because so far, out of 10,000 AI chatbots we've ever looked at, they've been the only AI chatbot that's been even close to being capable of challenging us on product quality. But if you work for this competitor, we would love for you to white label our stuff - I'm fairly confident in that you will realise now that competing with us is futile ... 😉

And yes, the bubbles theme our own AI chatbot is currently using was also blatantly stolen from you - Thx, I freakin' adore it! 😂

Two days of coding, now back to marketing and selling again 😊

Thomas Hansen

Thomas Hansen I am the CEO and Founder of AINIRO.IO, Ltd. I am a software developer with more than 25 years of experience. I write about Machine Learning, AI, and how to help organizations adopt said technologies. You can follow me on LinkedIn if you want to read more of what I write.

Published 13. Apr 2024