CodeFest 2017, Alexander Graebe (Uber Technologies Inc), Context First: What Context Platforms teach us about great experiences

https://2017.codefest.ru/lecture/1205

Stop thinking of your platform as a solution to improve user's life as a whole and focus instead on life contexts to create memorable user experiences. This talk will outline the concept of Context Platforms and the value of openness and extensibility.

What does Facebook, Alexa, Yelp, Airbnb, and Uber have in common? They all provide experiences that make your life easier and more enjoyable. We use platforms like these as part of our life with a very specific context in mind. At Uber, we call them Life Context Platforms. Great Context Platforms leverage a unique life context and optimize it for the best possible experience. However, the best Context Platforms also enable third-party applications to leverage that unique life context through an open and extensible ecosystem. These third-party applications, also called Agents, can be a stepping stone into a different life context. Agents allow connecting a variety of life contexts to shape the perfect, fluent user experience. For instance, a Facebook Agent for Uber would allow you to order an Uber while you connect with your friends. In your Uber, a Nest Agent could allow you to set your thermostat into economy mode. The Yelp Agent could show you reviews of the place you're headed to... That creates a magical experience!

Context Platform ecosystems serve a critical goal: Make your life context even more personal, and with that more valuable. You can choose your personal goals and infuse context information from other platforms to improve your life context experience.

Join me for a presentation about the concept of life contexts and how current platforms act as agents that fulfill your very specific, contextual goals. You will understand how life contexts should shape your experience to make your users have wow-moments!

Transcript

1. Context First What Context Platforms teach us about great experiences Alexander Graebe Developer Relations Lead Uber
2. Today is April 1st
3. Where we are headed Mobile First AI First Context First
4. domo.com
5. youtube.com Every online user maintains 2-5 social media accounts on average
6. wayray.com
7. nest.com
8. sas.com Data is still trapped in silos
9. xkcd.com
10. ~ 35.000 decisions every day
11. designorate.com Decision quality decreases
12. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Vestibulum vitae neque dolor. Duis erat diam, bibendum quis suscipit vel, lobortis vel mauris. Fusce mattis sit amet magna sit amet dictum. Aenean id sem a urna iaculis bibendum eget eget risus. Pellentesque tristique quam ac erat vulputate ultrices. Duis dictum blandit ipsum, dapibus finibus justo ornare ut. Alexander mauris eros, egestas eget unlearnt vitae, feugiat sollicitudin ante. Graebe vitae nunc sit amet erat porttitor rutrum sed et neque. Maecenas et feugiat orci, et mattis velit. Aliquam faucibus nisl to volutpat tincidunt. Maecenas tellus lorem, speaks non iaculis a, aliquam dictum velit. Curabitur facilisis felis eget magna volutpat posuere. Duis non nisi non erat imperdiet molestie. Russian nec ornare velit. Maecenas hendrerit quam ex, at euismod sem varius eget. Interdum et malesuada fames ac ante ipsum primis in faucibus. Nunc tincidunt nunc at pharetra venenatis. CodeFest ut massa vehicula, ornare libero vitae, iaculis ante. Aenean mattis elit eros, eu ultricies ex laoreet nec. Aliquam about. Suspendisse potenti. Context non massa mollis neque efficitur egestas. Nulla libero tellus, vehicula at dignissim hendrerit, Platforms in justo.
13. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Vestibulum vitae neque dolor. Duis erat diam, bibendum quis suscipit vel, lobortis vel mauris. Fusce mattis sit amet magna sit amet dictum. Aenean id sem a urna iaculis bibendum eget eget risus. Pellentesque tristique quam ac erat vulputate ultrices. Duis dictum blandit ipsum, dapibus finibus justo ornare ut. Alexander mauris eros, egestas eget unlearnt vitae, feugiat sollicitudin ante. Graebe vitae nunc sit amet erat porttitor rutrum sed et neque. Maecenas et feugiat orci, et mattis velit. Aliquam faucibus nisl to volutpat tincidunt. Maecenas tellus lorem, speaks non iaculis a, aliquam dictum velit. Curabitur facilisis felis eget magna volutpat posuere. Duis non nisi non erat imperdiet molestie. Russian nec ornare velit. Maecenas hendrerit quam ex, at euismod sem varius eget. Interdum et malesuada fames ac ante ipsum primis in faucibus. Nunc tincidunt nunc at pharetra venenatis. CodeFest ut massa vehicula, ornare libero vitae, iaculis ante. Aenean mattis elit eros, eu ultricies ex laoreet nec. Aliquam about. Suspendisse potenti. Context non massa mollis neque efficitur egestas. Nulla libero tellus, vehicula at dignissim hendrerit, Platforms in justo. Algorithms help understand data
14. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Vestibulum vitae neque dolor. Duis erat diam, bibendum quis suscipit vel, lobortis vel mauris. Fusce mattis sit amet magna sit amet dictum. Aenean id sem a urna iaculis bibendum eget eget risus. Pellentesque tristique quam ac erat vulputate ultrices. Duis dictum blandit ipsum, dapibus finibus justo ornare ut. Alexander mauris eros, egestas eget unlearnt vitae, feugiat sollicitudin ante. Graebe vitae nunc sit amet erat porttitor rutrum sed et neque. Maecenas et feugiat orci, et mattis velit. Aliquam faucibus nisl to volutpat tincidunt. Maecenas tellus lorem, speaks non iaculis a, aliquam dictum velit. Curabitur facilisis felis eget magna volutpat posuere. Duis non nisi non erat imperdiet molestie. Russian nec ornare velit. Maecenas hendrerit quam ex, at euismod sem varius eget. Interdum et malesuada fames ac ante ipsum primis in faucibus. Nunc tincidunt nunc at pharetra venenatis. CodeFest ut massa vehicula, ornare libero vitae, iaculis ante. Aenean mattis elit eros, eu ultricies ex laoreet nec. Aliquam about. Suspendisse potenti. Context non massa mollis neque efficitur egestas. Nulla libero tellus, vehicula at dignissim hendrerit, Platforms in justo. But, they create a filter bubble
15. We can do better with Context Platforms
16. Relax at home Catch up on news Prepare for meeting Navigate while driving Chat with friends Play video games See what friends do Search Carpool Manage household Manage project Date Commute with train Work out Watch videos Listen to music Search for parking Shopping Manage finances Find inspiration Life contexts indicate intent But, product interaction does not consider these contexts
17. Relax at home Catch up on news Prepare for meeting Navigate while driving Chat with friends Play video games See what friends do Search Carpool Manage household Manage project Date Commute with train Work out Watch videos Listen to music Search for parking Shopping Manage finances Find inspiration What works for
18. Relax at home Catch up on news Prepare for meeting Navigate while driving Chat with friends Play video games See what friends do Search Carpool Manage household Manage project Date Commute with train Work out Watch videos Listen to music Search for parking Shopping Manage finances Find inspiration doesn’t work for
19. Anticipate intent and make it visible audi.co.uk
21. Intent C Intent A Intent B
22. CONTEXT
23. USER STATE HTTP LOCATION BROWSER Legacy Contexts
24. platform: 'MacIntel', touch: 'enabled', timezone: '-420' Accept-Language: ‘en-US' latitude: '122.02394', longitude: ‘-45.294’ Legacy Contexts
25. Google Inbox: Delivery Context get delivery date and provide snooze option littlebigdetails.com
26. GitHub: Project Context get filename and provide LICENSE templates littlebigdetails.com
27. Google Photos: Life Context get photo metadata and provide search suggestion littlebigdetails.com
28. Why now?
29. 2009 ∑ Connected Things > ∑ Connected People wikipedia.com
30. 2012 Every atom on the earth could now be assigned an IP address ipv6ready.org
31. 2014 Users can now expose data in scopes securely wikipedia.com
32. 2014 1.2TB query < 10 secs posible amazon.com
33. 2015 Small, encapsulated services providing business value oreilly.com
34. 2015 Google open sources TensorFlow library for machine learning tensorflow.org
35. hugeinc.com 2016
36. 2017 > 200 real-time APIs are now available programmableweb.com only 1% of all APIs
37. Key concepts
38. User State All user preferences and profile details User interaction and current screen Account history and past transactions
39. Life Context Situation-specific user details categorized by life context Obtained in real-time through agents Should help determine user’s intent
40. User Intent Specifies user’s situational goal Defined set of intents is required Goal is to anticipate and suggest the right intent
41. Context Agent Agents bridge products Leverage user context and execute action Agent library maps to context library Instantiated through context trigger
42. Save on travel Context Agent Chatbots are one instance of context agents Anticipatory UX is an enabler Context agents are not limited to UIs
43. Anticipatory UX User-centric design discipline successfully leveraging machine learning to infer users’ preferences Key principles: 1) Simplified interaction 2) Min options for max use cases 3) Anticipating choices hugeinc.com
44. Everything should be made as simple as possible, but not simpler — Albert Einstein
45. Getting Started
46. Become context driven When building context driven products, keep in mind … hugeinc.com 1) Segment your user base 2) Reshape UX approach 3) Build smart algorithms 4) Build context architecture
47. Become context driven: Segmentation Understand your user segments and identify most relevant life contexts. Based on that, build a set of intents and find context platforms to get context details to anticipate these Commute Relax at home Shopping Chat with friends and relatives Find best transportation option Find best specialty store
48. hugeinc.com Become context driven: Anticipatory UX
49. Become context driven: Smart Algorithms Responsive: Allow users to teach algorithms using recurring feedback loops Transparent: Show users why an intent was anticipated Intelligent: Do not confuse action with intent Failsafe: Prepare for fallback in case agent doesn’t return relevant details on time
50. Your Product 3rd Party Product Life C ontext 3rd party can push context changes, user state unclear Become context driven: Context Architecture
51. Your Product 3rd Party Product Life C ontext id: 'foo@bar.com', time: 1315955008, type: 'checkin', venue: { … } Become context driven: Context Architecture 3rd party can push context changes, user state unclear
52. Your Product 3rd Party Product Become context driven: Context Architecture User state is clear, you can pull for context details
53. Your Product 3rd Party Product User state is clear, you can pull for context details user_id: 'foo@bar.com' latitude: '122.02394', longitude: ‘-45.294’, user_id: 'foo@bar.com' Become context driven: Context Architecture
54. Challenges
55. Challenges Data is paramount Real-time APIs needed Trust in algorithms and privacy Anticipate, don’t decide
56. The ubiquity of the Internet is […] reshaping not only how we interact with our devices but how they interact with us. — Aaron Shapiro
57. @agraebe Alexander Graebe Developer Relations Lead Uber Questions? agraebe@uber.com