Hyperautomation and AI/ML: A Strategy for Digital Transformation Success.pdf
IEN artificial intelligence, chatbots and virtual assistants for the enterprise - Brenton Charnley
1. LET’S DO
MORE THAN CHAT.
Brenton Charnley, CCO Flamingo.ai
16 May 2017
“Cognitive Virtual Assistants get the job done
and create value for both customer and
enterprise”
2. ABOUT FLAMINGO
Based in NYC and Sydney, Flamingo is an ASX listed Enterprise SaaS company, which
provides a Cognitive Virtual Assistant (AI) platform designed for automating online sales,
servicing and support. Flamingo’s Cognitive Virtual Assistant is called Rosie. Flamingo is one
of Gartner’s ‘Cool Vendors’ and clients include Fortune 50 financial services firms.
3.
4. 1. Chat – the new interface
2. It’s all about conversations
3. Chatbots v CVA’s
4. Creating more value with CVA’s
5. When to use a CVA
6. Are you ready?
7. What to look for in a vendor
8. Key takeaways
5. C o g n i t i v e V i r t u a l A s s i s t a n t s
( C VA ) g e t t h e j o b d o n e a n d
c r e a t e M O R E v a l u e f o r b o t h
c u s t o m e r a n d e n t e r p r i s e
19. • No longer just about digital
channel
• Start with the customer problem
first
• Help solve the problem digitally in
the channel used
• Better customer experience and
opportunity to delight
22. CHATBOT
“Are artificial
intelligence systems
that customers interact
automatically with via
text or voice”
COGNITIVE
VIRTUAL
ASSISTANT
“Are artificial intelligence
systems that can
understand natural
language and automatically
complete electronic tasks
for the customer and
delivers insights. ”
24. “Psychological processes
involved in acquisition and
understanding of knowledge,
formation of beliefs and
attitudes, and decision making
and problem solving. They are
distinct from emotional and
volitional processes involved in
wanting and intending.”
Source: Business Dictionary – “Cognitive”
/cognitive/
26. WHAT MAKES A BOT INTELLIGENT?
• Natural language processing
• Context in a process
• Decision making and action
• Learning and knowledge (new insights)
27. WHAT MAKES A BOT INTELLIGENT?
Source: The sense-think-act cycle in the context of robots (from http://makezine.com/2017/01/06/choose-use-sensors-robot/)
28. • Natural language
processing
• Learns instantly - no need
for large-scale programming
activities
• Has context – tight
integration with business
process
• Right decision and right
action – to get the job done,
now
29. No one goes to
a website to
just have a
chat…they
have a job to
do
38. LOWER COSTS HIGHER BRAND
ADVOCACY
MORE THAN JUST A CHAT
HIGHER SALES
CONVERSION
HIGHER CUSTOMER
SATISFACTION &
BETTER EXPERIENCE
39. HELPING SOLVE PROBLEMS
Conversational Commerce and ChatBots: Business & Consumer Usage and Attitudes Market Research by Fifth Quadrant, Commissioned by Flamingo & Sugar CRM , 2017
40. GETTING THE JOB DONE
Conversational Commerce and ChatBots: Business & Consumer Usage and Attitudes Market Research by Fifth Quadrant, Commissioned by Flamingo & Sugar CRM, 2017
41. Conversational Commerce and ChatBots: Business & Consumer Usage and Attitudes Market Research by Fifth Quadrant, Commissioned by Flamingo & Sugar CRM, 2017
42. The future is already
here, it’s just not evenly
distributed yet.
William Gibson
45. 1. Speed – immediacy
2. Destination – customer purpose
3. Complexity – data v workflow
4. Context – integration with
business process
46. DESTINATION
• What is the customer’s
destination?
• Enterprise and
customer aligned to get
the job done
• How hard is it to get the
customer there now?
49. USE CASES IN FINANCIAL SERVICES
QUOTE APPLY BUY OWN USE
CUSTOMER SALES ASSISTANTS
SERVICE ASSIST
CLAIMS ASSIST
INFO
FAQ ASSIST
AGENT SALES ASSISTANTS
AGENT SUPPORT
INSURANCE VALUE CHAIN
54. IMPACT OF OUR CVA TO DATE
30PT
INCREASE IN NPS
300%
DECREASE IN CALL HANDLING
TIMES COMPARED TO VOICE
20%
INCREASE IN CUSTOMER RETENTION
32%
INCREASE IN ENQUIRY
TO QUOTE
50%
REDUCTION IN
ABANDONMENT RATES
90%
INCREASE ON YOY
CUSTOMER RENEWALS
61. ENTERPRI
SE
SECURITY
WHAT TO LOOK FOR IN A VENDOR
FAST
IMPLEMENTATION
ACTUAL
MACHINE
LEARNING
SMART HUMAN
HANDOFF
API
READY
INSIGHTS &
ANALYTICS
CONFIGURATIO
N
NOT
CUSTOMISATIO
N
63. YOU NEED
MORE THAN A
BOT
STRATEGY –
GET READY
BOTS & CVA’S
CAN LIVE &
(WORK)
TOGETHER
Key Takeaways
PUT THE HUMAN
BACK IN THE
HUMAN
GO FOR THE
UPSIDE NOT
JUST
AUTOMATION
YOU NEED A
PARTNER TO
DELIVER
Good morning
Thank you
This morning I am going to discuss topic of CVA’s and chatbots and how you can drive real commercial value for your enterprise
Introduce Flamingo
Introduce role, Chief Commercial Officer which space across CFO/CMO/Head of Sales – pretty much anything that’s not technical, that’s for out data science and technology team.
Recently completed a trip across Asia meeting with over 30 insurers in 5 countries
There was a mixed response between AI, chatbots and machine learning
Has anyone else experienced this?
Somewhere between terminator is coming
They are going to take all of our jobs!
There is a real fear about AI in a lot
What is it and how does it work
Buzzword
Some are already trying chatbots, with some limited results and some are looking at more intelligent CVA’s
No longer just about digital channel
Interface has changed - CUI
Start with the customer problem first
Help solve the problem digitally in the channel used
Better customer experience
In my view, CVA’s can get the job done
And Create Value
Importantly, chatbots and CVA’s can both add value
For the customer – faster, more convenient interactions, getting the job done faster
For the enterprise – increased sales, customer experience and brand advocacy
Firstly
Importantly, solving customer problems have not changed, only the interface has
As we all know chatbots have been an increasing trend
Conversational Commerce a new term, delivered to the world by Chris Messina
2016 was year of bot strategy
2017 is the start of execution with more bots and less so more CVA’s coming to the fore
But we’ve been here before
We’ve been here before…
Customer experience poor
Low online sales
High abandonment rates – the empty shopping cart
No longer just about digital channel
Interface has changed - CUI
Start with the customer problem first
Help solve the problem digitally in the channel used
Better customer experience
No longer just about digital channel
Interface has changed - CUI
Start with the customer problem first
Help solve the problem digitally in the channel used
Better customer experience
Pretty much any bot you talk to is actually intelligent because it is receiving a command from you and responding to that command based on some logic.
This is because its goal is simply to respond to you and it is showing agency in simply responding
Robot – simple and repeated tasks
Have been used in automation for along time, manufacturing as example
Program based
Whereas, Cognitive Virtual Assistants need a human response
Need to be able to perform a task
Learning based
1. CVA can create value because of the interation of intelligence, automation and chat (link back to thesis)
Natural language processing – ability to understand, words, sentences and sentiment
Context in business process – where am I and what does this mean – “can you tell me more about this”
Decision making and action – what is the next step
Learning – to be able to remember, and provide new insights
Not just command and action, such as in rules based
Ability to be able to do this cycle is what drives the customer value
Coming back to me previous story and examples, need help at the point in time online, when I have the problem
I need help at Question 4 or Step 7, and I need you to know what my questions means in context of where I am
At that point in time
Sense
Need to understand where it is, sense the environment,
How we apply this, where am I in a business process or problem
Decision
Based on that and data input
Make a decision about what to do next
For us, the next step is critical as this is what helps to progress the customer
Increase solving problems and in particular sales
Act
Need to be able to take action
Integration with business process here
Critical to enable the full automation
Meet the time requirement for the customer
Natural language processing
Can be deployed quickly
Learns instantly - no need for large-scale programming activities
Has context – tight integration with business process
Right decision and right action – to get the job done
Remembers and optimizes process
The key difference between a chatbot and a CVA
Ability to be able to get the job done
And this is where chatbots often fall down
This is how I see some chatbots, they aren’t smart
Never ending questions because they are rules based
Don’t have the context or intelligence to make the next step for the customer
Heard from Amir from Slack yesterday
Simple bots can be deployed without Natural Language processing that can add great commercical value
Simple bots can create value too
Subway bot
Rules based, prompts, limited questioning
Use of bot platforms such as API.AI and Chatfuel are great examples of how easy it is to build rules based chatbots
So there are plenty of bots that an create value
so why are CVA’s better at it than bots?
So why can CVA’s create more value
Today, going to a website is like trying to navigate what and then figuring out how to get there
We have all experienced this,
Just like my optus story from the start, easy to get lost, and what if I don’t know where to go
Countless ways to get there, but really, you just want to start
And if anything like me and my childhood, lots of arguments between my parents as to who took the wrong direction
This is the current experience for most people in digital channels
Recent pitch I did to a Investment Retirement Fund in Australia
On their website was a “useful” search bar that said “ask us anything”
Now, I thought this was great:
I could set my destination
Wouldn’t have to click through/,
I was mistaken
Key words search
Resulted in 247 different response
None of which were helpful
Using the map analogy, is see CVA’s can solve this problem
CVA’s to some extent can do what google maps can do for us
Start with the destination in mind
Think about best way to go
Take in requirements
Then guide me through what I need to do
Re-configure if I need to change
This is about personalisation
That is, get me to where I want to go, faster and easier, and make it relevant to me
Do I want to take the bus, train, walk or car, give me options and help me to avoid the roadblocks
Today, people want personalised, they want a better experience
These benefits split between automation and upside
But, we see that the upside is important
Some of these benefits are
Lower costs
Must have a good ROI
Can’t forget the costs
Business case usual question
Customer experience
Consistency
Multiple conversation at the same time
New insights
Better experience leads to
Higher Brand Advocacy
You can delight the customer
Ultimately, this adds to driving better commerce
Bots are getting better conversion rates compared to traditional channels such as email and direct marketing campaigns
We get to help the customer at the point in time they need, with the right information
link back to optus example and thesis
Both chatbots and CVA’s about timing
Help me at the point in time I need help
The current reasons why enterprise aren’t solving these problems online
Surprisingly, I am intrigued by this one
Current online solutions don’t work very well
Organisation readiness
Change
Knowledge
Work with early adopters
To Scott Bair’s comment yesterday, there are followers, but not necessarily fast
For example the base use case right now
To solve customer problems that require a human response…quickly
Break this down into four categories
Speed is first and paramount to automation – that is think about now, email, delayed but human response, voice similar delay, timing is crtical
Destination
Complexity
Context
Will talk through each of these further
Complex workflows might mean many difference types of questions
Similarly for product information, if complex then a CVA could assist by answering questions at that right time
Importantly one thin we have learnt here that better to deploy multiple CVA’s, rather than one complex CVA
Finally context
Again does your customer expect you to know specific about where they are in a process
Details like name, step in a process
Know what is the next step
So if you combine all these you can get some really interesting use case
Where we are at the moment, here are some use cases
Mention use cases
We are working with a number of larger fortune 50 insurers across Australia and US
For example, Nationwide Insurance is one large US insurer to be trialling our AI in direct to consumer sales
Replacing booking a meeting with guided sales tool rather than use agents
Replacing static forms with conversations
Replacing static FAQ’s with intelligent conversations and then guide the customer to the next step
This can also be used for knowledge management for employees as an example
Some of our results to date
Demonstrate impact of use of CVA’s
No you know the impact and some use cases
Next question is are you ready?
2016 year of the strategy, but execution is hard part and 2017 is start of this
Do you have the team in place
“Organisations don’t innovate, people do”
Do you need a head of chat? Argue
Create space for a Bots in your organisations
Should be connected directly with business process
It isn’t just innovation, need operations to come along and connect with sales and marketing
Pentagon of pain – need a team to be assigned to make it happen (cross discipline team)
Do you have the digital maturity for chat?
Are your people ready?
Do you have the data available and infrastructure to support it
Do your customers need or want chatbots or CVA’s?
Integration key priority / API / linkage to business processes
Key to getting to full automation
Biggest barrier when working with companies today is the lack of integration, so get started on this one
Automation v Upside
If you are using the words business case together with innovation, sometimes it might not get across the line
This links back to the strategy, get the right strategy and buy in
Chat interfaces may not be suitable for all businesses and your digital maturity
Get the buy in
So if you are ready and you know that you want to use a CVA
What to look for
Industry knowledge – templates, learning, data, insights
Actual machine learning – lots of people say they are, but are they? Get under the hood and understand how it learns. We fuse Bayesian analysis with neural nets - a hybrid strategy in our approach.
Enterprise security – One of the key issues, cybersecurity is paramount, trust is a big issue for consumers. Where we are in sales and financial services, Insurtech more specifically, if you get it wrong, you get it wrong big, also not regulation
Fast implementation – you are looking to experiment, so try quickly and try different use cases, therefore need to be quick
Smart Human – don’t forget the humans, as learning happens need to learn from human/customer interaction
Configuration – customisation takes a long time
API ready
Insights & Analytics – can the platform help you to improve the journey and discover new insights
So now you know
how they differ
Some use cases on how you can apply
How to get yourself ready for CVA’s
Summarise a few of the takeaways
Bots and CVA’s can add value,
but CVA’s can add more because of the nature of complex human problems,
There is a huge upside for changing the customer experience – just think of how low conversion rates are for email and web conversion
We
More specifically
Bots & CVA’s can live together
Both can add value
Find the right use case
B2B
You need more than a Bot Strategy – you need to be ready
Execution paramount
Find the right use cases
Go for the upside
There is a great play here for changing the customer experience
New product development
New insights
ROI not just on efficiency
Put the humans back in the human
Automation enables you to