2. what are chatbots?
Programs that use
conversation as
interface
Programs that
function on a top of
messaging platform
“Chatbot is also known as
chatterbot, talkbot, bot, chatterbox,
or Artificial Conversational Entity”
4. @galiyawarrier
“Classic” chatbots
1960s-70s
ELIZA (Doctor)
a simulation of a psychotherapist
PARRY
a simulation of a person with
paranoid schizophrenia.
chatbots
history
page
04
Alan Turing
1950
Paper on the topic of AI
“Computing Machinery and Intelligence”
Introduces Turing test concept
Considers a question
“Can machines think?”
5. @galiyawarrier
chatbots
history
page
05
1990s – 00s
“General” chatbots
Research using NLP-backed chatbots continues
Albert One, Alice, Cleverbot,
many others
They become more and more mainstream
Wikipedia runs ~2000 bots
Applications within automated online assistance,
call centers, dating / romance apps,
conversational toys, etc.Loebner Prize: annual AI competition based on
Turing Test
Malicious bots on IM applications
6. @galiyawarrier
2016
Facebook launches bot
platform for Messenger as F8
with a bot engine hosted by Wit.ai (April),
allows to take payments (September)
Google announces Allo & Assistant services at Google I/O (May),
launches Allo and acquires API.ai - a conversational platform (September)
Kik opens a bot store
chatbots
history
page
06
Microsoft’s rough start with Tay chatbot,
announces Bot Framework
at //Build (March)
Slack, WeChat (China) and others
continue to use bots
Oracle presents its chatbot
platform targeted at enterprise
architectures (September)
8. @galiyawarrier
why
now?
page
08
Users are tired of searching for and installing
yet-another-mobile-application
Advanced natural language processing tools
available for developers via API calls.
Mobile app developers struggle to make their
products visible to the right audience
9. of consumer’s time on smartphone
spent using 5 apps*85%
* Each user have their own selection of those five, but mostly games, social networking or messaging apps
Source: https://techcrunch.com/2015/06/22/consumers-spend-85-of-time-on-smartphones-in-apps-but-only-5-apps-see-heavy-use/
10. @galiyawarrier
benefits
for business
page
010
They can be where their customers are:
inside those messaging applications
nick dutch
head of marketing @ domino pizza uk
“Essentially we are simply
adapting to our customers’
behaviours rather than trying
to target new customers.”
Source: https://www.marketingweek.com/2016/08/17/dominos-pizza-explains-why-it-is-chasing-facebook-bots/
https://medium.com/chris-messina/2016-will-be-the-year-of-conversational-commerce-1586e85e3991#.8r6oolxlu
chris messina
developer experience lead @ uber
“…utilizing chat, messaging, or other interfaces
to interact with people, brands, or services …
The net result is that you and I will be talking to
brands and companies over FB Messenger,
WhatsApp, Telegram, Slack, and elsewhere
before year’s end, and will find it normal.”
11. @galiyawarrier
benefits
for you as developers
page
011
Build once
Users already on
chat platforms
Use on different
channels
Integrate additional
capabilities
(APIs: NLP, search, etc.)
12. @galiyawarrier
benefits
for bots
page
012
Available anytime and anywhere
Can interact with humans in multiple ways
Text, UI, Voice options depending which is more preferred in a specific situation
Can deal with complex or mundane (for humans) tasks
Process large sets of information; Finding patterns; Talk back-and-forth
Consistent way of interacting with various services
13. @galiyawarrier
bot
patterns
page
013
01
Transactions
Lookup, reference and information
seeking scenarios backed by a data
source
• “Book me 2 tickets for Deepwater
Horizon movie using my credit card for
tomorrow evening”
03
Advisory role
Prescriptive guidance via “expert
systems” based on user input:
• “How much I can save if I switch to a
payment plan X?”
02
Social conversations
Ability to sense sentiment and
engage in open-ended conversation
within bot area of “expertise”
• “Your product is horrible. Is there anyone I
can talk to about it?”
04
Information retrieval
Lookup, reference and information
seeking scenarios backed by a data
source:
• “What are the features of service plan B?”
16. @galiyawarrier
bot stack
(survey)
Survey: “What’s in your bot stack?”
40 responses, 19/08/16 - 29/08/16
(multiple responses allowed)
Source: https://aka.ms/bot-survey
The results of the survey were far less definitive…The answers to the various
forms of “I want to build Bot Type X, what should I use?” are very much TBD at
the moment
There’s a lot of flexibility if you’re currently getting started building bots. The
frameworks and language tools accessible to anyone with an interest in
spinning up a bot on one (or many) platforms is incredible, considering that
most of the major advances and releases have been in the past 18mo.”
page
016
18. @galiyawarrier
microsoft
bot framework
page
018
Bot Builder SDKs
Build great dialogs
within your Node.js- or
C#-based bot
01
Bot Connector
Connect your bot(s) to
text, Skype, Slack, and
other services
02
Bot Directory
Try a bot and add it to
your conversations
03
19. @galiyawarrier
• Create bot using C# or node.js
• You need to host it somewhere
• SDK supports dialogs to model a conversation:
• Reusable
• Persistent dialog state
• Various dialog types
• SDK allows for rich interactions:
• Rich attachments (image, card, doc, video
• Support for calling (Skype*)
• Chat Emulator
bot builder sdk
(.net, node.js, rest)
Your conversation logic
Logic
Web Service Cognitive Services:
LUIS, Translation, etc.
25. @galiyawarrier
Bot Framework
Emulator
• Send requests and receive responses to/from
your bot endpoint on localhost
• Inspect the JSON response
• Emulate a specific user and/or conversation
Download from:
https://aka.ms/msbot-emulator
*
27. @galiyawarrier
Cognitive services:
LUIS
page
027
(aka Language Understanding)
• Determines intent and detect entities
• Seamless integration with Speech
Recognition
• Learns over time
• Use pre-built models from Bing &
Cortana
• Models work across devices
Your bot will interact with LUIS by routing queries to LUIS API
for intent and entity matching
30. @galiyawarrier
bots for
everything?
page
030
noyes
Tasks with lots of (sequential) options
(i.e. appeal for parking tickets)
Expected conversations
(i.e. order a pizza, book an appointment)
Non-time critical queries with many
options (i.e. book restaurant for team
off-site)
Help on a singular task with many variables
(i.e. find fastest route from A to B)
Time critical requests
(i.e. book restaurant / taxi for now)
No, not all interactions are conversations!
31. @galiyawarrier
challenges,
opportunities
page
031
chatbots still in their infancy
Bots are simple, may be developed out of curiosity and
hype, people can play with them and then abandon.
lack of universal platform / standards
Bots have neither standards nor mature technology stacks
(compared to web’s W3C)
bot discovery still an issue
Consolidated “app store” for bots?
33. @galiyawarrier
additional notes,
links and resources
page
033
• https://chatbotsmagazine.com - Chatbots magazine
• http://conferences.oreilly.com/artificial-intelligence/bot-ca - The first-ever O'Reilly Bot Day
• https://botness.fuselabs.org - Botness Gathering / Unconference (06/2016)
• https://aka.ms/msbot-resources - (All) Microsoft Bot Framework Resources
• http://www.donotpay.co.uk/signup.php - Lawyer Bot
• https://www.producthunt.com/tech/ask-elon-musk-bot - Chat one-on-one with Elon Musk
P.S. Title of talk was inspired by a quote from Dave Barry, a Pulitzer Prize winning author and columnist
35. thank you!
Galiya Warrier
Cloud Solution Architect (Azure)
Microsoft UK
galiyawarrier
github.com/Galiya
Slides are available at:
https://aka.ms/chatbots-dddnorth