AI AI AI AI AI AI AI AI AI AI AI AI
AI AI AI AI AI AI AI AI AI AI AI AI
AI AI AI AI AI AI AI AI AI AI AI AI
AI AI AI AI AI AI AI AI AI AI AI AI
AI AI AI AI AI AI AI AI AI AI AI AI
AI AI AI AI AI AI AI AI AI AI AI AI
AI AI AI AI AI AI AI AI AI AI AI AI
AI AI AI AI AI AI AI AI AI AI AI AI
Slide 4
Slide 4 text
Intro to AI
UX patterns
Wrap-up
Q&A
Slide 5
Slide 5 text
No content
Slide 6
Slide 6 text
No content
Slide 7
Slide 7 text
No content
Slide 8
Slide 8 text
No content
Slide 9
Slide 9 text
No content
Slide 10
Slide 10 text
No content
Slide 11
Slide 11 text
No content
Slide 12
Slide 12 text
No content
Slide 13
Slide 13 text
No content
Slide 14
Slide 14 text
No content
Slide 15
Slide 15 text
No content
Slide 16
Slide 16 text
No content
Slide 17
Slide 17 text
Would you sell
this?
Would you buy
this?
Can we productise
AI technologies
today?
Slide 18
Slide 18 text
What are AI technologies?
Language models
Image models
Audio models
Video models
…
Various types of input
⬇
Weights, embeddings
⬇
Various types of output
Slide 19
Slide 19 text
What can AI technologies do?
Generate text
Generate images
Generate video
Generate audio
Manipulate text, images,
video, audio
Recognise patterns
Scrape content
Transform content
Translate content
Analyse natural language
…
Slide 20
Slide 20 text
How do we productise?
What do we need?
Slide 21
Slide 21 text
Where is the industry now?
Gold rush
Models are available, on-premises and cloud
Some models are open-source
Everyone and anyone can use them
Not much technical expertise necessary
You can build something in a couple of days
Slide 22
Slide 22 text
No content
Slide 23
Slide 23 text
nhost.io
Slide 24
Slide 24 text
R&D
Expertise
Infrastructure
Cost
UX
Not a blocker
Not a blocker
Not a blocker
Not a blocker
???
Slide 25
Slide 25 text
R&D
Expertise
Infrastructure
Cost
UX
Not a blocker
Not a blocker
Not a blocker
Not a blocker
expectations, behaviours, beliefs,
motivations, attitudes, preferences,
patterns, conventions,…
Slide 26
Slide 26 text
AI and user experience
Slide 27
Slide 27 text
How do users use AI
technologies?
Slide 28
Slide 28 text
No content
Slide 29
Slide 29 text
Let
’
s ask the experts
Slide 30
Slide 30 text
No content
Slide 31
Slide 31 text
We don
’
t have distinct
“AI patterns”
Yet
I think there will be
Slide 32
Slide 32 text
No content
Slide 33
Slide 33 text
Would you talk to AI in
daily life?
If your entire o
ff i
ce
suite is point-and-
click?
Slide 34
Slide 34 text
Real products need to sit
into existing tool
landscape
We interact with AI
computers just like other
computers
Slide 35
Slide 35 text
UX patterns
Miscellaneous collection, observed by me
Nothing academic or official
Slide 36
Slide 36 text
Text
prompt
• Very common
• Microsoft Copilot, Midjourney,
Notion
• Similar to Spotlight, Google
search
Slide 37
Slide 37 text
No content
Slide 38
Slide 38 text
Chatbot
• Long sessions with back-and-forth
• Expectation of memory and
iteration
• Amazon, ChatGPT, Gemini,
Observable AI assist
Slide 39
Slide 39 text
Access to
custom data ☝
Slide 40
Slide 40 text
🐣
Slide 41
Slide 41 text
“Google Maps
launches a new
AI” or “new
chatbot”
Slide 42
Slide 42 text
Image
input
• Where do I get the image?
• Live camera UX is different
from file upload
• We already have patterns for
image input
Slide 43
Slide 43 text
Image to
text to image
https://reads.alibaba.com/ai-
image-tools-ecommerce/
Voice
• Quick responses
• Clean environment
• Need to concentrate
• Can
’
t glance information
• How to follow up?
• Privacy and security?
Slide 49
Slide 49 text
VR, video, gestures
Slide 50
Slide 50 text
No content
Slide 51
Slide 51 text
No content
Slide 52
Slide 52 text
VR, video, gestures
• VR has been around for a while – still difficult
• We have knowledge about the UX patterns: how to
avoid nausea, enable precision, interpret input
• Needs specialised equipment
• Libraries and APIs exist
• Keeps failing on the market
Slide 53
Slide 53 text
Emerging UX patterns
My emerging vocabulary specific to generative models
Slide 54
Slide 54 text
“Prompt-output”
Slide 55
Slide 55 text
“I sketch,
you
fi
nalize”
sketchlogo.ai
Midjourney
Slide 56
Slide 56 text
“You sketch,
I
fi
nalize”
➡ ???
https://reads.alibaba.com/ai-image-tools-ecommerce/
Slide 57
Slide 57 text
“Press for
magic”
Slide 58
Slide 58 text
“Press for
magic”
🪄 Remove background
Slide 59
Slide 59 text
“Just do it”
Slide 60
Slide 60 text
“Inform
without
input”
Slide 61
Slide 61 text
“No nothing”
Slide 62
Slide 62 text
What do all of these
have in common?
Slide 63
Slide 63 text
Text prompt
Chatbot
Image input
Image to text to image
Voice
VR
Video
Gestures
“Prompt-output”
“I sketch, you
fi
nalize”
“You sketch, I
fi
nalize”
“Press for magic”
“Just do it”
“Inform without
input”
“No nothing”
Slide 64
Slide 64 text
https://www.samsung.com/latin_en/support/mobile-devices/how-to-use-live-translate-for-phone-
calls-on-the-galaxy-s24/#:~:text=Once%20you%20have%20set%20up,Tap%20on%20Live%20translate.
https://www.youtube.com/watch?v=6LOY4KHzFOQ
Check consumer
expectations
Many “AI products” are
live for consumers
already
Slide 65
Slide 65 text
User expectations are
high
Error tolerance is low
Slide 66
Slide 66 text
Imperfections and slow
responses can be UX deal
breakers
AI does not excel at these
Slide 67
Slide 67 text
“Trust but
verify”
Slide 68
Slide 68 text
No content
Slide 69
Slide 69 text
Productisation
Slide 70
Slide 70 text
R&D
Training
Infrastructure
Cost
UX
Not a blocker
Not a blocker
Not a blocker
Not a blocker
???
Slide 71
Slide 71 text
R&D
Training
Infrastructure
Cost
UX
Not a blocker
Not a blocker
Not a blocker
Not a blocker
Not a blocker
Slide 72
Slide 72 text
You can productise AI
models right now
Slide 73
Slide 73 text
R&D
Training
Infrastructure
Cost
UX
Use cases
Jobs to be done
Value
Slide 74
Slide 74 text
Can we create
something totally
novel?
Slide 75
Slide 75 text
Of course. But it
’
s very
di
ff
i
cult
And you will probably
fail many times, like
most innovators
Slide 76
Slide 76 text
Full “AI work
fl
ows” will
still take a while
Start with clearly
de
fi
ned speci
fi
c tasks
Slide 77
Slide 77 text
You can manipulate bitmap images
right now using image models
Can we manipulate 3D
fi
les with
these models?
Maybe not with these exact
models, but there
’
s a reference.
Something to build on
Slide 78
Slide 78 text
What can AI technologies do?
Generate text
Generate images
Generate video
Generate audio
Manipulate text, images,
video, audio
Recognise patterns
Scrape content
Transform content
Translate content
Analyse natural language
…