Home > Columns
Cognigy Takes Conversational AI to the Next Level
Click image to download the eBook
Cognigy.AI offers an Enterprise Conversational Automation
Platform for customer and employee service automation. Available in both
on-Premise and SaaS, Cognigy.AI enables enterprises to connect to their users
on any conversational channel, including chatbots, virtual assistants or voice bots
on phone lines. The graphical toolset allows both non-technical users and
developers to build and manage complex interactions that go far beyond simple
FAQs towards automating complex business processes. The company had been
recognized in 2018 in “Gartners’ Cool Vendors in AI for Conversational
In our series of conversations with suppliers that were
scheduled to exhibit at Enterprise Connect last month before it was converted
into a virtual event, CrmXchange Managing Partner had an in-depth
(non-automated) conversation with Derek Roberti, VP Technology, North America
Let’s start with an overview of your platform.
The challenge organizations face today is one of volume. The
volume of customers has not diminished. Across many industries it has
increased. At same time, the number of resources available to service those
inquiries, the agents, has gone down. This is due to the changes that had to be
made last year. It’s become harder to recruit people who are willing and able
to work in agent positions. It’s even more difficult to retain them. This has
created a perfect storm in the contact center universe; the mismatch between
demand on the customer side and the resources to resolve all their issues.
There are simply not enough agents to handle the repetitive
tasks of updating data and providing easily accessible information. This
includes inquiries that a customer should be able to do for themselves using
self-service tools. It had become a luxury for businesses to have agents
performing rote activities that don’t require human judgment or empathy. This
also plays into the ability of companies to retain agents. Taking tasks that
are both boring and prone to human error off their plates and giving them more
enriching work allows companies to better retain agents. The enterprises can
then focus on supporting them with the right tools and giving them more
Cognigy provides the technology that enables companies to
automate voice and chat-based conversations and complete end-to-end business
processes on those channels. Non-developers are able to quickly become more
productive, building out automated conversations in less time and at lower cost
than using traditional IVRs.
Have there been any studies that have looked at agent
retention and determined if the work is more interesting for them?
At this point, there are no metrics in place to reliably
measure retention. The metrics that do exist gauge deflection rates and
decreased handle time. Now, much of the
front end of a conversation has been done before it reaches the agent. While
there are many ways to evaluate productivity of agents or the volume of calls that
they take, we still can only try to interpolate the effects on retention. There
is an ongoing dialogue around how automation technologies impact human
resources: does it take away the role of a person or does it augment their
value? How does it all fit together?
Does the automation replace or augment the agent?
If you put yourself in the position of an agent who takes
calls to look up an individual’s information and repeat it back to them, that’s
a 3-minute conversation. The next one and the next one is the same thing.
Because it is so repetitive and unvarying, that’s nowhere near as challenging
as it is mentally exhausting. When we are working with clients who are trying
to automate certain parts of the business process, the entire contact center
team down to the agent level is always involved. They are the ones who most
want repetitive categories of work automated. While it is of course driven by
their desire to make their work more interesting, it is also fueled by their
passion for delivering meaningful customer service. They feel like it is not an
effective use of their time to perform interactions that don’t require the
One way that we materially contribute is streamlining the
front end of the conversation. Often, a large part of the agent interaction is
verifying that the customer is who they say they are. We can take that burden
off the agent’s shoulders.
What are your first steps when a company automates?
When we start working with a company, we ask them what are
the top 10 or even the top 100 questions that come into the contact center. We
then triage which ones that can be most easily automated. That would be project
1. Other companies may identify a niche
problem. While it may not be the organization’s highest volume issue, it’s a
certain kind of end-to-end business process that can be implemented with
conversational AI. We build that out as
a pilot phase and then it is automated. The folks who know the processes best
are the ones who perform these activities every day, so enlisting their help is
essential. They know both the stated process and the actual process as it plays
out in conversations with customers.
Can chatbots be personalized?
We often say that voice bots and chatbots are as smart as
the systems and data to which they have access. We want to have the mechanics
in place to be able to fulfill users’ requests. But the other side of that is
having the ability to personalize the interaction while making it as fast as
possible. One example of how that works would be some of the major airlines who
instantly know who I am when I call based on my phone number. Depending on the industry,
some businesses might want to employ a second basis of information. It could be
the last 4 digits of the social security number it could be zip code or flight
number. With that level of verification, the business can not only say they
know who the customer is but ask if they are calling about their flight that
was just cancelled. This not only still surprises some customers that a system
can work that well but demonstrates to them that the company cares about them
enough to invest in technology that improves their customer experience.
One mistake that we still see is that when a customer is
passed on to a human agent, the agent is required to repeat that verification
process. This is not only an unnecessary activity for the agent, but
disappointing from the customer perspective they think “Hey, didn’t I just do
that with the machine and now I have to repeat it with the agent? It is quite
important for companies to figure out how it all works together to provide a
seamless end-to-end customer experience.
We are talking about inbound interactions at this point.
What about chatbots for outbound calling?
While the technology to deliver such experiences is
available for outbound, many companies haven’t yet thought ahead in terms of
the user experience they want to design. What we see in outbound is companies
which want to qualify lead generation: where people may have filled out a form
on a website and are then called to verify information, at which point they are
passed on to a human agent. This is the most common automated outbound scenario
at this point. We can see other
applications in the future, such as flight rescheduling or a healthcare
provider setting up appointments for flu shots. The trickiest part of outbound
automation is the trust level of consumers who already are wary because of the
volume of dubious calls they receive.
Perhaps one of the biggest advantages of conversational AI
is that it is multi-channel in its most effective manifestations. We can have
conversations that go from voice to SMS to webchat on a website and have that
be a seamless experience. One area that can be credible and successful is
SMS-based notifications. We’re able to take those text messages and make them
conversational either through having a dialogue with an SMS channel or by
bringing them into a webchat experience or even by having a user request a call
to them. A business could start an automated conversation in that fashion.
When you look at things in the multi-channel context, what
we are trying to do is optimize things for the user by keeping them in the
self-service mode and help them as much as possible. This creates a
competitively strong experience. Many of the things you will see in this space are
legacy players that are focused on the voice channel and newer players who are
optimized for digital channels. Our technology ensures that these channels work
together seamlessly without needing two separate products to enable it or not
to have it be double the work with one product because you have to design
different voice and digital experiences. These elements need to be considered
What differentiates Cognigy?
Cognigy is a low-code or no-code platform for building out
automated conversations. What does that mean? In the traditional mode of IVR
creation, business have two options. They either pay the vendor for the professional
services to make whatever IVR menu or functional changes necessary. The other
is that they have someone in their organization who can figure out the tooling
that their vendor provides. In either case, it’s not fast and it requires using
outside resources and often necessitates having skill sets that the contact
center doesn’t have unless they are able to lean on internal IT. The vision of
low-code or no-code platforms is to provide tools that a non-technical or
non-developer person could use to iterate quickly instead of making changes and
then having them implemented immediately instead of a month later. This
empowers the non-technical personnel to design and evolve and improve automated
conversations at will.
Cognigy puts the tooling in the hands of the subject matter
experts who know the business processes and the information they need to share with
users—and do everything possible to empower them to use these tools to create
automated conversations without having to knock on IT’s doors or incur
professional services fees. We’re moving into the era of what is called the citizen
developer where non-technical user has the power to build applications on their
own using non-coding approaches.
There are many conversational AI products on the market. What are some of the differences?
In the conversational AI market, what you’ll see on the
lower end are genuinely easy-to-use tools that are usually employed for a
specific purpose. For example, if a business wants to do lead capture on its
website, it might have a simple wizard-based tool that enables it to click on
someone’s name and email address to schedule a message, appointment, or
callback. On the other end of the spectrum, there will be tools that enable
organizations to do just about anything. But the catch is they will need heavy
IT involvement to accomplish anything.
What Cognigy provides is an offering that serves the needs
of audiences with both simple and complex application requirements. It also
enables them to get tasks done quickly with very facile user interfaces that
allow them to push the complexity of conversations to a point farther than
other tools could do without the need to write code.
Regardless of who is doing the implementations, they can be
as much as 10 times more productive accomplishing it with our platform as
opposed to traditional coding-based approaches. The power of our platform has not
only been validated by our customers who have compared us with every other
significant vendor in the marketplace, but also by Gartner and other analysts
who see that low-code capability as our sweet spot.
Who in the organizations do you find is responsible for
pulling everything together?
The people responsible for making it work are those in
customer experience roles who can live in different parts of the organization.
Some are in customer support focused on serving customers while other companies
look at it more broadly, incorporating customer acquisition as well. The
business problem that most businesses look for us to resolve is the need to be
more efficient. This has accelerated within the past 10 months or so as they
find that they either can’t hire people fast enough or at sufficiently affordable
rates to handle the volume they are experiencing. They know it’s imperative to
reduce hold times to improve their NPS scores and frequently, it is the contact
center VP who approaches us to say that they’ve got to get a handle on what is
happening, and the ownership lies in terms of how well they do.
Sometimes we encounter experts within the companies
themselves who tell us that they know conversational AI is part of the roadmap
for the future of customer service within the organization. Thus, they want to
start developing dedicated resources who can become experts in deploying this
When I am out in the world attempting to do business as a
consumer, many of the chatbots I see are basically glorified search engines…and
not particularly good ones. There are several reasons for this, and I believe
the most obvious one is a failure of companies to develop a vision on how they
are designing these conversations from a user experience perspective. But I
also can attribute the problem to many companies seeing the end result as saying,
“How can we take the take the traditional FAQ on our web page and put it into a
The more appropriate perspective would be for them to
realize that chatbots are most powerful in every dimension when they can be
used to solve end-to-end problems for customers. Having the right API
integrations is essential for creating the ability to deliver this more
Cognigy has many out-of-the-box pre-built integrations with
leading enterprise-grade technologies. What’s more important is how we created
these extensions using our proprietary extension framework which is a blueprint
for how to build extensions into applications. We thought about how companies
often have legacy or homegrown systems that have no available published APIs,
yet we still need to be able to integrate with them. If you build such APIs
using our blueprint, they become surfaced in our graphical low-code editor as
though they were native parts of the platform. The impetus is to write the code
once and give the non-developer access to the integration within the Cognigy
We also have out-of-the-box channel integrations with
everything from WhatsApp and Facebook Messenger to webchats such as those a
company would have on its website to SMS, voice and all of the most widely used
channels. We can automate conversations on internal channels such as Slack. But
we can’t predict everything out there and sometimes, in such areas as mobile
apps, there’s no one standard for how a specific one is going to communicate
with an external application. So, we’ve built in the ability for a company to
create its own channels. We call these Transformers, which is our technical
name for it, but we make the platform extensible in that ways as well. As an
example, a major insurance company which is a long-time customer of ours,
wanted to integrate Cognigy into their mobile app. They didn’t have to change
anything in their mobile app to be able to connect and they were able to
accomplish it due to the ability for a business to either create its own end
points or customize our out-of-the-box end points.
What sort of questions do you ask people as they stop by
When we are at a live event, usually the customers who come
to our booth are thinking about automating some aspects of their customer
service. One of the first things we try to determine is what kind of money are
they trying to save and what problems are they trying to crack. Many times they
want to increase their contact center efficiency by handling use cases that
don’t require human decision making. Sometimes it’s getting them to understand
that they can resolve a gap where their competitors are able to provide more
personalized experience. In other cases, it’s finding out what their pain
points are that automation can solve for them. We also want to learn what
channels they are looking to automate: voice or text or both. Surprisingly,
some companies are even new to live chat, much less automated chat. But what we
really want to do is help them think through the business problems that
automation can solve for them.
The domain of conversational AI is new to many contact
center operators. While immersed in data in such areas as agent productivity,
average handle time and NPS, they might not be aware of some of the
conversational analytics that come out of interactions with automated bots,
whether voice or chat.
You have a new feature, Insights. Can you tell us about that?
We have introduced what we consider a unique feature this
year called “Insights” which is a way to glean knowledge that can’t be produced
by traditional analytics in contact centers with human agents. We create funnel
metrics that tell businesses how many people started interactions with a bot
and what was the nature of their inquiry. And of those, how many created this
end-to-end business process. This allows them to see where any drop-offs came
in the process and how they can optimize the information to create a better
user experience. They can categorize conversations and resolution types to see
what reasons people need to speak to a human agent or determine if there are
any ways that the automations are not working. This gives them empirical means
to measure the effectiveness of their investment and have the means to make
better, data-driven decisions to improve their customer and employee