Frequently, once brands decide they need a bot they’ll pick the platform straight away. That’s before they’ve even considered the use case.
Generally speaking, their thinking goes like this; ‘Google’s one of the biggest tech companies so we must succeed if we use their Cloud NLU’, or ‘we use Microsoft for everything else – why shouldn’t we use their LUIS service’, or’ we’ll just go with Amazon’s AWS to start and swap it later if required’.
As Christoph Börner argues, there’s nothing wrong with any of those providers, but each of them has their strengths so you need to pick them depending on your goals. Christoph is CEO of Botium, and he explains a much better approach to Kane Simms in his VUX World interview.
It seemed like a good idea at the time
Botium is a provider of bot testing tools. They focus on the testing, training and monitoring of bots.
However there’s another service Botium provides that’s a game-changer – technology benchmark testing. As far as conversational AI goes, it’s a brilliant way to save a lot of wasted time and money because you start the process with testing.
It goes like this; you start with your training data, intents and utterances. Then Botium test the various systems and decides what’s the best option for you, considering your domain and training data.
So instead of starting with a platform, building your bot and then finally testing it to see how well it performs, you start with testing to check how the platform performs before you build it!
It shouldn’t be about copying what other people do, or going with your preferred partner, or any other arbitrary reason. Your domain and training data are unique to you, and therefore your NLU (Natural Language Understanding) should also be selected uniquely for your needs. As Christoph says “there are domains where you would be super surprised how one [NLU] is outperforming the other.”
The best place to start is to approach Botium with “what you have and what you want to build, and let’s try it out.”
Start with the right mindset
As Christoph says “The biggest [knowledge] gap I see within our clients is knowing the needs and requirements of their end users.”
It’s all about the users. A bot that’s not designed for the end users is destined for failure. In Botium’s experience, customers are most keen on bots that provide accurate answers, a great user experience, data privacy and performance.
Remember, bots are not websites or touchscreen apps. The approaches that worked for other ‘traditional’ software development need a serious upgrade. When you tested apps in the past you probably didn’t have to consider the different regional accents of your users for example. Now that’s one of the first things you’ll need to consider when you create a conversational experience.
Consider enlisting Botium at the start of your bot project before you’ve selected any other provider. Don’t rush into the wrong choices early on, expecting to change it later. As Christoph says “changing at this [later] point will be a vision of hell.”
Best make the right choice before you see those gates!
You can hear the full interview with Christoph here.
This article was written by Benjamin McCulloch. Ben is a freelance conversation designer and an expert in audio production. He has a decade of experience crafting natural sounding dialogue: recording, editing and directing voice talent in the studio. Some of his work includes dialogue editing for Philips’ ‘Breathless Choir’ series of commercials, a Cannes Pharma Grand-Prix winner; leading teams in localizing voices for Fortune 100 clients like Microsoft, as well as sound design and music composition for video games and film.