Bot Primer !
It's a Gold Rush
What to Expect ?
History & Introduction
Rules to Build better bots
Platforms to Build
My favourite bots
BOTS
1988
Jabberwacky
One of the earliest attempt on creating an AI through human interaction with an ultimate goal of moving from text-based system to wholly voice operated
1972
Parry
Parry was more serious and advanced than Eliza - it was described as "Eliza with attitude"
MIT prof. Joseph Weizenbaum developed, the first ever chatbot, that could “interact” with people like a psychotherapist would. He used a script that would recognize certain patterns and keywords, and generate a response accordingly.
1966
Eliza
H
I
S
T
O
R
Y
2001
Smarterchild
An intelligent bot widely distributed across SMS network .with features such as quick data access & personalised conversation .it was considered as precursor to Apples Siri & Samsung's voice
1995
A.L.I.C.E
"Artificial Linguistic Internet Computer Entity " ALICE was a natural language processing bot.
AI Speech synthesis program created for MS - DOS- based PC's . Designed to showcase a digitazied voice
1992
Dr. Sbaitso
H
I
S
T
O
R
Y
2012
Google Now
Developed by google for google search mobile app , it employs natural language UI to answer questions , make recommendation & perform action over web.
2010
SIRI
Intelligent personal assistant , it uses natural language UI to answer questions and perform web service request.
Watson uses natrual language processing & Machine learning to reveal insghts from large amount of data
2006
IBM-Watson
H
I
S
T
O
R
Y
2016
Bots-Messenger
Facebook launched bots for messenger platform which allows developer to create bots that can interact with FB user .
2015
Cortana
Intelligent personnel assistant crafted by Microsoft cortana sets reminder ,recognises voice commands & answer questions .
Voice service inhabiting the amazon echo device , alexa's capable of voice interaction - she used Natural Language processing algorithms to receive , recognise, and respond to voice commands .
2015
Alexa
H
I
S
T
O
R
Y
Let’s get something out of the way
i love bots
most bots are dumb
some bots are powerful
If I asked everyone here what a “bot” was
you would all have different answers !
A bot is any software application that lives in a messaging tool.
It doesn’t need to be a text based interface.
It just needs to live in a messaging platform.
Rise of Messaging Apps
Rise of Messaging Apps
Let’s look at some strategies for building
powerful bots instead of dumb ones.
12
Rules
to build better bots
RULE
bots are about platfrom
#1
Whether it’s Telegram, WeChat, Facebook, Slack or
another platform, the power of bots is that they
LIVE IN A TOOL WE ALREADY USE
RULE
look to the future
#2
THIS IS HOW YOU BOOK A DOCTOR IN WECHAT
It doesn’t look like most of the “bots” we see today
WeChat has a BILLION users. their platform is used like crazy.
This has 27 buttons
This has 2 buttons
Interface comparison
RULE
context is powerfull
#3
Building on a platform means that you know a huge amount about your users on day one.
Narendra Modi
Age : 57 years old
Gender : Male
From :Guj , IN
Slack tells you about how they work
Facebook tells you about who they are
Twillio now lets you learn about who they are
RULE
bots are about platfrom
4
Meekan is a bot that helps schedule meetings.
If it were an iPhone app, everyone on the
team would need to install it.
Instead, it lives in #Slack, where it has access
to your team by default.
The Slack platform enables new kinds of apps
Meekan
“Meekan schedule a team meeting tomorrow at noon”
TAKE A STEP BACK
Why are we talking about bots? Two words.
VENTURE CAPITALIST
A bunch of VCs decided to give money to bot companies
and now everyone needs a darn bot strategy.
This is the Gartner hype cycle at work.
Tech comes in waves…
Before bots we had
SOCIAL NETWORK
Before bots we had
MOBILE
APPS
Before bots we had
MOBILE SOCIAL APPS
then we got
MOBILE SOCIAL VIDEO APPS
There was one week in 2011 when 11 "Instagram for Video " apps were launched
These apps all launched in 2011, Only one snapchat
is making a kill with VC's , rest are DEAD
That is to say,
Get ready for a deluge of bots
RULE
it does not have to be mass market
#5
In a crowded market one winning strategy
is to build niche apps
You only need a small percentage of FB
Messenger’s 1 billion users
to build a great business
RULE #6
Structure thing people are already are doing
THERE ARE SO
MANY PROCESSES
THAT CAN BE TURNED INTO BOTS
Why try to
create a
new behavior ?
DON'T INVENT THE NEW BEHAVIOR
Profit it
1
Find existing behavior
Bot it
2
3
RULE #7
redefine conversatations
Howdy.ai is a great bot
it take synchronous meeting & parallelizes them
Howdy.ai is a great bot
it take synchronous meeting & parallelizes them
Someone on the team triggers howdy.
Howdy goes out and ask everyone a question.
Howdy turn the answers into a simple report.
1
2
3
just because a or process used to happen one way , doesn't mean computer's can't restructure it .
RULE #8
avoid disasters , use human to filter
Blossom bot is the #1 best bot of all time
* Machine Learning
It was built in house at the New York Times.
It uses *ML to identify stories that are prime to go viral.
Stories it finds get 3.4x more views.
Blossom bot is the #1 best bot of all time
Blossom could connect to Twitter.
"Instead, it connects to Slack where the social media
editors can have a conversation about what makes
sense for each of their 100+ social media accounts."
RULE #9
Eliminate Decisions
Imagine two ways to order a pizza
imagine two ways to order pizza
Pizza App one
Super custom
Customize volume of each topping
Pick your cheese types
imagine two ways to order pizza
Push a button
Get a Pizza
Pizza App Two
WHICH MAKES A BETTER BOT ?
Curate to avoid decisions
RULE #10
build guard rails
Even Simple bot can go wary
If your bot feels like it can answer anything,
people will ask it anything.
DON’T MAKE IT FEEL LIKE IT
CAN ANSWER ANYTHING
RULE
understand your platform
#11
Build experiences that makes sense for the platform you're on.
Target their audience and speak their language .
understand your platform
A Tale of Two Bots
Helps teens deal with
anxiety and depression
Helps journalists find
breaking news
Lives in Kik
Lives in Slack
RULE #12
this is still new. no one knows what to expect
Platforms
Discord
Telegram
Slack
Facebook Messenger
Kick
- Octane AI
- wit.ai [bought by Facebook]
- howdy’s botkit
- api.ai
- textit.in
- Motion.ai
- Chatfuel
- IBM’s Watson
- BeepBoopHQ
- Dexter [owned by Betaworks ]
- converse.ai
- Gupshup