Creating the conversational web
Chat needs to be disrupted. It has become too difficult for businesses to deliver an acceptable experience, which is degrading user perceptions of it’s value. Left unchecked, customers are just going to stop using it and potentially one of the highest converting website tools will be gone. For a while it seemed chatbots might be a potential solution, and there was a surge of growth over the last few years. Based on the latest data from Segment.com and Zoominfo, that exciting growth is slowing and most companies don’t understand why.
In a perfect world, brands would be able to put knowledgeable, engaged staff on chat around the clock, but this isn’t a perfect world. As a result, customers are left with unresolved questions and frustrating experiences. To fix this, we needto disrupt the status quo of chat, chatbots, and standard websites by making it possible for our websites to be interactive the way our televisions, smart speakers, and other various devices have become. If I can ask a smart speaker to turn on the lights, or play a song, it’s time we make it possible for customers to ask our websites the question they want addressed, and simply get an answer.
Introducing the Conversational Web
Gamalon was founded to pioneer the next generation of AI. From our beginnings at DARPA, the focus has been to make AI accessible and usable by anyone. Our success creating explainable, agile AI has made it possible to provide any size business a conversational AI experience for their website - one that is fundamentally different from current cookie-cutter chat solutions. Gamalon’s vision is to provide an always on, instant response mechanism on every page of a website - an AI assistant integrated with the website that actually knows details about that company’s products and services (and can recognize the question when asked.) This vision for a conversational website isn’t just adding a live chat function to a website, but rather integrating conversational AI deeply into the web experience such that it becomes clear you can just ask the website what you want to know - and it will tell you.
This required the Gamalon team to completely rethink what it means to create a conversational experience on a website, examining successful and unsuccessful designs in conversational technologies and charting a path to an experience that is intuitive and comfortable, while indicating that this is a fundamentally different experience from other chat and chatbot solutions.
Alternative to General AI
A conversational website isn’t a general AI solution - we don’t think it needs to tell your customers the airspeed of a pigeon or the weather in Albuquerque. Instead, our vision is a service that enables any potential customer on your website to ask, “Which product can help with managing meeting room bookings” or “Does your application integrate with Hubspot?” or “how do i upgrade the room displays to work with Office 365?”
And rather than simply route customers to FAQs or help portals and force them to figure out your information architecture, a conversational website should handle 80% or more of sales support questions to be handled instantly, providing an experience that suggests your customers come first.
Rewriting Mobile and Web Chat
To start, we started designing an experience that felt integrated into the website but didn’t feel like a separate chat window. We wanted it to feel like you were having an SMS conversation with the website. It was important for the experience to feel more like a “conversation” and less like a sales or support interaction. It was also important for it to simply look different from all the happy robot chat windows out there.
A critical part of getting a conversational web experience right was to fix the unfortunate mobile experience. We decided to fundamentally change the mobile web experience, enabling visitors to get immediate answers to questions without the extra clicks and awkward navigation on mobile websites. Mobile users have voted with their thumbs, and text has become the dominant communication medium on mobile devices. Unlike mobile web navigation, which often gets spread out across the 4 corners of a mobile device - the keyboard is conveniently located and users have PLENTY of muscle memory about using it.
Why is this important? If you have ever tried to chat on a mobile device, the first thing that happens is it opens a new tab and separates you from the website. Neither digital marketers, nor consumers like this approach. It forces separation from the content you are there to see. By contrast, a conversational website creates a seamless mobile chat experience that feels like part of the page.
Beyond Website Content
A conversational website should know what your website knows. Product and feature information, use cases and pricing - these are table stakes. FAQs are called FAQs for a reason, so a conversational website must be able to answer these. And since all of this content already exists, a marketer shouldn’t have to recreate all of it. To accomplish the team at Gamalon had to invent conversational content management so that it would not be fragmented and distributed across decision trees and rules engines.
The “IdeaBase” concept is a first of it’s kind solution - a simple to use interface designed for business users like marketers to see what the AI knows about the company’s products, features, use cases, FAQs, and more. Because Gamalon is explainable AI and not a black box, it makes it possible for the team to organize the ontology of knowledge within the AI and visualize it so that marketers can see what the system knows, and what it doesn’t - making it easy to update gaps in knowledge and edit existing knowledge simply through the product interface.
Previously, updating a machine learning model required labeling a significant amount of training data - something most companies have neither the time nor resources to handle. Instead, Gamalon learns by reading a company’s website, ingesting product documents like a product catalog, and converting those into conversational models. From there, every conversation teaches the system new things and in the event an answer isn’t available in the IdeaBase the AI alerts the appropriate business users about the knowledge gap and allows them to provide an answer in real time.
As products and features change, updates are as easy as navigating to the right field in the interface and changing the description, or adding a new integration and description of how it works. Finally your product marketing team has complete control over how customers questions are answered about products, even the day they are launched.
The best part - you never have to answer the same question twice. This means 30 days after launch, the system will know anything that was asked that it didn’t know on day 1. Same on day 90. Like an employee that never forgets and doesn’t need to be told twice, Gamlon becomes the employee you always wanted answering all your front line sales and support questions online.
One thing chatbots do well is lead the visitor to a goal - typically a request for contact information. It was critical that our AI understood it had goals too, primarily the goal is to figure out why the customer is visiting and then help them accomplish the goal.
Have a support question? Gamalon will work to understand the customer need, clarifying if necessary, and then confirm the chat was helpful.
Researching the right product to solve your needs? Gamalon will ask enough questions to be able to recommend the right product and see if it makes sense to escalate to sales.
It’s not enough to answer questions, a good website leads the customer to the right information to solve their problem, and a conversational website has to do the same.
Something we have spent a lot of time on, but only scratched the surface on, is trying to decide if the AI engagement should feel like a human interaction.. Our current focus is on making it “human-ish”. Human-ish means it’s ok if the customer knows it’s a bot, we just have to make it good enough that they don’t care. People don’t visit business websites to strike up a conversation and discuss art - they need information. If the most efficient way to get the right information from a business was carrier pigeon, we’d all have a flock of them. For the most part, all we want is to solve our problem.
Here is an example: If you called the support line for your bank and your options were to have your question answered immediately by AI, or wait 5 minutes to talk to a person - the majority of people will choose the immediate answer. The key is that people want a system that allows them to just act human, not have to figure out the special keywords necessary to get an answer. The complaints about Alexa, Siri, and Google Assistant are not that they are a little robotic when they respond, but that they don’t sufficiently answer their questions.sers simply want the right answers to their questions, and fast.
This is what it means to be human-ish.
Learns Over Time
There are so many more features that make a conversational website, but one of the most important is that we made it so the system learns over time. This was a common refrain from every conversation I’ve had about AI tools and chatbots - people feel like they don’t learn anything. Just like an employee, a conversational website needs to be able to adapt to new questions, easily consume new content, and ask for help when it doesn’t know something. This is one of the most innovative aspects of Gamalon that makes the conversational web possible - a workflow and process where unknown questions alert users and enable them to easily add answers in real time. Something we’re incredibly pleased with is that we made it so that each new question trains the AI, so you never have to answer the same question twice.
Over time, a conversational website feels more like a versatile, knowledgeable rep that never sleeps more than it does a website. For brands with lots of content, a conversational website is a way to leverage that content more than just blog posts and tweets.
Gamalon is excited to introduce the most significant innovation in conversational technology since Siri, and we are looking forward to the age of the conversational web.