Anypoint Code Builder , Google Pub sub connector and MuleSoft RPA
Bots: The Unspoken Challenge of Conversations
1. •
•FIN. DIGITAL
•PAGE # • PRESENTATION BY: MARCUS & RAKIA
Bots: The Unspoken
Challenge of
Conversations
we are UMA
Marcus Finley & Steve Milner
TechXLR8
2. • The State of Bots
• Bots in Action
• Bot Development Strategy
• Challenges of Bot Development
• Designing Conversations
• Enabling Speech-to-Text
• Design Intents & Responses
• Case Studies
What we’ll discuss
It’s a lot…we know
3. SAFE SPACE
THE STATE OF BOTS
Botsarethenewapps.
ConversationsarethenewUI.
AIistheprotocol.
Messagingappsarethenewbrowsers.
5. SAFE SPACE THE STATE OF BOTS
• 90%
ofourtimeonmobile
isspentonemailand
messagingplatforms
• 16%
OfAmericanjobs
AIwillreplaceby
theendofthe
decade.
(Forrester)
6. Machine Learning (ML) vs Artificial Intelligence (AI)
• Machine learning and AI are not the same.
• AI is when a machine mimics “cognitive” functions
• ML is teaching machines to make decisions in
situations they have never seen.
•The most mainstream approach to ML is showing
the algorithm of a data set of situations and telling it
what the right decision is — training a model.
Buzzwords
7. Natural Language Processing (NLP)
•AI must communicate with humans as well as
humans communicate with each other.
•NLP is a huge priority and challenge in AI
research.
•Because human communication is not
straightforward. It’s a complex web and it
depends hugely on context.
Buzzwords
11. Rules-based bot:
•Can only respond to very specific commands
•Only as smart as you program it to be
Machine Learning/AI bots:
•Understanding language and not just commands
•Continuously gets smarter as it learns from
conversations
Hybrid Bots:
•Combination of programmed responses but
machine learning is applied.
Bot Family
19. 1. Identify the purpose of your bot
•Platform selections
•Limitation of platforms
2. Create intents
3. Integrate services
4. Train
5. Test
6. Deploy
Creating a Strategy
20. To learn most language all you need is 300
hundred words… that’s not enough for bots.
• What makes a bot valuable & sticky?
• Speech-to-text
▪ Languages
▪ Names
• Intent recognition
▪ Handle jargon
▪ Unforeseen intents
▪ Context
Challenges with Bots
21. But before we even get to vocabulary…
Challenges with Bots
Can you say my name?
26. • Like a human assistant, an AI assistant can only
be as smart as the information — the context —
you give it access to
• Context is king when it comes to complex tasks
• Predicting conservation pattern is essential
• Error handling
▪ What do you when the bot doesn’t know
what to say?
Context, Context, Context
28. • Define the context (location, time, etc.)
• Define bot role (instructor, concierge, advisor, etc.)
• User Journey Mapping
• Desired emotional connection (encouraging,
inspiring, empowering, etc.)
• Research other bot design patterns for solutions
• Write out the initial script of the conversations
Conversational UX Tips
30. • Native device handling (ie - iOS)
• Amazon Polly
▪ 34 different voice options
▪ Example of Inflections: Voice 1 & Voice 2
• Lyrebird (Under Development)
▪ Ability to copy a human voice and replicate
▪ Example: Voice Sample
Levels of Voice Quality
31. • Amazon Lex
• Azure bot tooling: https://www.luis.ai
• Google Cloud Speech (currently using)
▪ Asynchronous voice string to decrease the
response time
▪ Provide both voice and text response as
separate strings
STT Service for Bots
38. SAFE SPACE
Q&A
BOTS ARE FUN - TRY IT!
Sign-up for a Google Home giveaway at: http://vyper.io/c/1211
39. FIN. DIGITAL
Resources
Microsoft - Bot Development Platforms
https://www.producthunt.com/posts/microsoft-bot-framework-2
Bot Builder
https://www.motion.ai/?ref=producthunt
https://smooch.io/integrations/motion-ai/?ref=producthunt
Facebook Messenger Bot
▪ https://octaneai.com/?ref=producthunt
Image Recognition AI
▪ https://www.producthunt.com/posts/vize-ai-custom-vision-api
Learning more on bots and AI
▪ http://aireads.top/?ref=producthunt