1) AI is currently experiencing a "big AI Spring" due to improvements in data availability, processing power, and interfaces that have increased data for training models.
2) However, there is also significant hype around AI capabilities that often misrepresent the current state of the technology. AI systems require specific, high-quality data and focused problems to solve in order to deliver real value.
3) The speaker advocates focusing on using AI to empower employees and improve customer experiences, rather than replacing humans, in order to realize transformational benefits while managing expectations.
3. Hi, I’m Rick.
● Co-founder & CEO of Guru
● Previously co-founder of Boomi
● Wish I pursued career in music
● Instead, I work on difficult enterprise
software challenges*
*Almost as exciting as being a rockstar...
Rick Nucci
Co-founder & CEO @ Guru
getguru.com
6. ML
CS
AI
NLP
DL
Incorporating human intelligence into
machines
Making machines "understand" the
meaning of natural language
Algorithms making machines learn from
data
Deep layered neural network: set of ML
algorithms inspired by the structure of the
biological brain
What is Artificial Intelligence?
18. We see lots of jargon and confusing terms
“Makes use of machine
learning, deep learning and
transfer learning to build a
unique Answer Graph”
“We train a deep neural network model by
converting historical customer service
transcripts into numerical representations
called word vectors”
“Do more with your data:
AI for professionals”
“AI Delivered
(by AI)”
getguru.com
22. The new religion
of artificial
intelligence is
called Way of
the Future.
“You will be able
to talk to God.
Literally. And
know that it’s
listening.”
getguru.com
29. getguru.com
“Artificial Intelligence in Business Gets Real”
● It’s early days, but real value is already being seen by companies
big and small
● It is being applied to revenue generating opportunities,
not “cost savings” opportunities
● It is being leveraged everywhere (show the graphic of supply
chain, sales, service, etc.)
30. What have we learned
from building Guru?
getguru.com
34. We realized that our customers
build daily habits around Guru
The DAU/MAU Ratio is a widely-used metric to understand product adoption. Best-in-class
consumer products—like Facebook in 2012—have a DAU/MAU Ratio around 60%.
*Source: Sequoia Capital
Typical enterprise
software*
Well-adopted
enterprise software*
10%
20%
41%
45%
60%
35. Artificial Intelligence
Guru’s unique approach of being “where you work”
enables collection of high-value, solution-specific training
data to power AI capabilities.
● How, when, and where the extension is used
● What Cards are searched for, viewed, used
Solutions
Foundation for content creation,
discovery and integration.
● Browser Extension
● Cards
● Verification Engine
● Slack Bot
● Content Creation & Management
● Analytic Dashboards
Core Infrastructure
Machine Learning powers Guru specific solutions,
such as AI Suggest and Contextual Coaching Capabilities.
● Suggest accurate & relevant information (unstructured data)
● Process guidance with 100% accuracy (structured data)
The Guru Cycle
37. getguru.com
AI is only as good as the data it
can learn from.
Focus on solving specific business problems with access to
valuable data to train from.
39. AI can cause CSAT and
agent satisfaction to drop
“Customer satisfaction levels will drop as companies drive more
traffic to chatbots, self-service, and chat that are not fully optimized
to engage customers effectively. As companies look to increase
customer engagement on digital lofty goals are being set particularly in
the area of call deflection. Some companies have goals of decreasing call
volumes by more than 50% in just under two years.”
Source: Forrester, Nov 2017 “Predictions 2018: Blended AI Will Disrupt Your Customer Service And Sales Strategy
43. Product usage generates training data which powers AI-driven suggestions
and predictions. The user takes action on these suggestions, which then
needs to feed back into the model.
Training Data → AI-driven Insights
→ User Actions → Better Data
45. AI is HARD.
You could spend forever building the perfect model. It is
better to start with a basic model and refine it over time
with your customers.
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
Remove the copy
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
Use steve’s version
AI according to wikipedia: “a system that perceives its environment and takes actions that maximize its chances of success". Artificial intelligence is a broad field, which includes many other subfields (again, using Wikipedia definition): reasoning, knowledge, planning, learning, natural language processing (communication), perception and the ability to move and manipulate objects.
Deep learning is one of many approaches to machine learning. Other approaches include decision tree learning, inductive logic programming, clustering, reinforcement learning, and Bayesian networks, among others.https://www.kdnuggets.com/2015/01/deep-learning-explanation-what-how-why.html
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
Illo
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
Bottom graphics, make as large as possible
Bottom graphics, make as large as possible
Add - heading back in - we see articles like this
Clean up
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
Anthony Levandowski, the first AI Religion. https://www.wired.com/story/anthony-levandowski-artificial-intelligence-religion/
Most promising ML technique today is ML, and specifically DL - which is in the hype phase
AI might be at “Frog” levels today, but it’s improving rapidly and much faster than we can comprehend.
AI might be at “Frog” levels today, but it’s improving rapidly and much faster than we can comprehend.
AI might be at “Frog” levels today, but it’s improving rapidly and much faster than we can comprehend.
AI might be at “Frog” levels today, but it’s improving rapidly and much faster than we can comprehend.
AI might be at “Frog” levels today, but it’s improving rapidly and much faster than we can comprehend.
so now that we know perhaps a bit about ai, what should we look for when evaluating solutions? how do we cut through the noise? here are some considerations.
First, the market opp
We continue to see Customer Service industry undergoing a massive shift. Companies are starting to realize that good service differentiates companies, and that the human connection between customer service teams and their customers is directly correlated to revenue; upsell, freemium conversion, repeat purchasing, customer loyalty
In 5 years customer service will shift from a Cost Center to a Revenue Center
Despite this, most new tech in AI is being designed to automate and replace the customer support rep. It flies in this face of this very real trend!
So this felt like a huge opportunity for us
First, the market opp
We continue to see Customer Service industry undergoing a massive shift. Companies are starting to realize that good service differentiates companies, and that the human connection between customer service teams and their customers is directly correlated to revenue; upsell, freemium conversion, repeat purchasing, customer loyalty
In 5 years customer service will shift from a Cost Center to a Revenue Center
Despite this, most new tech in AI is being designed to automate and replace the customer support rep. It flies in this face of this very real trend!
So this felt like a huge opportunity for us
Second the data
We started tracking our engagement data, to understand if our customers have sustained adoption w our product. This is the #1 problem w products in this category, no one uses them!
We saw very strong data, w 45% of our users using guru every day
But we dug deeper...we looked at where they use guru
What is the use case?
We found that of our users, the strongest engagement is happening with our customer support teams
We also found that they are doing very specific things...solving tickets and chats with Guru
Guru is augmenting their workflow and empowering them to solve customer problems faster, more accurately, with confidence
We spent the next year proving the concept, building the AI infrastructure, working with customers, launching
To successfully increase the efficiency of the Care Department you have already identified the installation of a robust knowledge management network, with innovative technologies, as the solve to achieve these business outcomes
ml is only as good as the data it trains on. beware solutions that claim to do too much! the more narrowly focused the solution is, the faster it will learn. just like humans right? most of us aren’t doctors, lawyers, engineers, teachers and pilots right? we tend to pick a field or 2 and go deep.
ml is only as good as the data it trains on. beware solutions that claim to do too much! the more narrowly focused the solution is, the faster it will learn. just like humans right? most of us aren’t doctors, lawyers, engineers, teachers and pilots right? we tend to pick a field or 2 and go deep.
ml is only as good as the data it trains on. beware solutions that claim to do too much! the more narrowly focused the solution is, the faster it will learn. just like humans right? most of us aren’t doctors, lawyers, engineers, teachers and pilots right? we tend to pick a field or 2 and go deep.
ml is only as good as the data it trains on. beware solutions that claim to do too much! the more narrowly focused the solution is, the faster it will learn. just like humans right? most of us aren’t doctors, lawyers, engineers, teachers and pilots right? we tend to pick a field or 2 and go deep.
your team is your collective intelligence. your agents, your content specialists, and subject matter experts. they all need to be part of the AI feedback loop for it to work. Avoid the temptations around quick wins, as they will quickly become quick loses. think about how much faster knowledge changes these days? we build, ship, and improve products today faster than ever before. the knowledge management process needs to move just as fast.
remember the last time you had an amazing support experience? remember the last time you had a shitty one? how about your support team?
your team is your collective intelligence. your agents, your content specialists, and subject matter experts. they all need to be part of the AI feedback loop for it to work. Avoid the temptations around quick wins, as they will quickly become quick loses. think about how much faster knowledge changes these days? we build, ship, and improve products today faster than ever before. the knowledge management process needs to move just as fast.
so now that we know perhaps a bit about ai, what should we look for when evaluating solutions? how do we cut through the noise? here are some considerations.
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy
Combine last two slides
i am not hear to talk about specific products or solutions focused on ai
there are just too many!
instead, i am here to give you a lay of the land around ai. what the heck does it mean, where it can help, and what to look for looking at your ai strategy