Allen Lu

1 followers · 1 following · 296 views

on the atlas · 10

highlights · 62

In an agentic world, you don’t write the AI's exact responses. Instead, you prompt the agent on how to generate a response.
Conversation Design Process Explained: Key Insights
What if there’s an error and the agent can’t access current weather information?
Conversation Design Process Explained: Key Insights
helping you test recovery paths and edge cases
Conversation Design Process Explained: Key Insights
serves as the foundation for a more detailed map.
Conversation Design Process Explained: Key Insights
Keep in mind the key that you defined in the SPEAKING model and sketch out the basic back-and-forth of the interaction.
Conversation Design Process Explained: Key Insights
Your job is to define the context and guardrails for the AI, giving it the tools and rules to create its own responses that are still aligned with the brand and the user’s needs.
Understanding Conversation Design for Better Relationships
You provide the agent with a persona, context, and a set of prompts and instructions. You set the rules for what the agent must and mustn’t say. Instead of writing the specific response, you write the guidelines for the agent.
Understanding Conversation Design for Better Relationships
this partnership is guided by a concept called the cooperative principle
Understanding Conversation Design for Better Relationships
Data-driven systems require designers to establish frameworks for learning and create mechanisms for appropriate adaptation. For users, this shift means interfaces become more personalized but potentially less predictable, creating new challenges for establishing trust and setting appropriate expectations.
What Is AI-Powered UX? course lesson | Uxcel
Traditional interfaces rely on explicit rules created by designers: "If users click X, then show Y." These rules remain static until manually updated by developers.
What Is AI-Powered UX? course lesson | Uxcel
When explaining AI to stakeholders, distinguish it from automation by emphasizing its ability to improve without explicit reprogramming.
What Is AI-Powered UX? course lesson | Uxcel
While an automated email system sends identical messages based on triggers, an AI-powered system might analyze response rates, customize content based on user behavior, and continuously refine its approach.
What Is AI-Powered UX? course lesson | Uxcel
These interfaces also need special attention to accessibility.
Conversational UX & Multimodal Interfaces course lesson | Uxcel
Users might speak a command, see a visual response, and then type their next request.
Conversational UX & Multimodal Interfaces course lesson | Uxcel
Chat interfaces work differently than voice systems, and visual AI has its own approach.
Conversational UX & Multimodal Interfaces course lesson | Uxcel
left side of a page the most
Ontario.ca style guide | ontario.ca
people do not read web pages word for word.
Ontario.ca style guide | ontario.ca
If your user does not need the content to complete their goal, leave it out.
Ontario.ca style guide | ontario.ca
Your ontario.ca audience could include: residents looking to perform a task, such as applying for an OHIP card or a hunting licence businesses looking for guidance about Ontario labour laws farmers looking for information about crops or business support parents looking for information about the school curriculum municipal employees looking for information about laws or regulations
Ontario.ca style guide | ontario.ca
If something is not covered here, follow the Canadian Press style.
Ontario.ca style guide | ontario.ca
When this does happen, support users through the transition.
Service design playbook | ontario.ca
Establishing continuous service improvement cycles
Service design playbook | ontario.ca
In many ways, live involves repeating discovery, alpha and beta in smaller iterations.
Service design playbook | ontario.ca
User’s needs and behaviours change over time, so services must adapt to keep pace. User research, testing and analytics should be used to make continuous improvements.
Service design playbook | ontario.ca
Without proper resources devoted to live, services become quickly outdated and fail to meet both the Digital Service Standard and user needs.
Service design playbook | ontario.ca
While most people understand the purpose of live, it’s not always given the attention and resources it deserves.
Service design playbook | ontario.ca
update any required policies to support the service
Service design playbook | ontario.ca
At a minimum, analyze service usage with Google Analytics.
Service design playbook | ontario.ca
This cycle repeats itself until all the features in the backlog are complete.
Service design playbook | ontario.ca
Before each sprint, the team selects features from the backlog, determines how to implement them and assigns the work to someone. Before work begins, the team must understand the objective of the sprint.
Service design playbook | ontario.ca
The research and prototyping from discovery and alpha help to form a prioritized list (backlog) of user stories. This backlog functions as a list of features and improvements to incorporate into the service.
Service design playbook | ontario.ca
If any of the above apply to your project, it might be good to go back to discovery or start alpha over.
Service design playbook | ontario.ca
Developers, subject matter experts and content designers
Service design playbook | ontario.ca
decide - the team selects a single option to move on to testing
Service design playbook | ontario.ca
Not only does this lead to better design outcomes but it also helps to ensure that the team has a common understanding of the product and any design decisions that the team has made.
Service design playbook | ontario.ca
Prototypes are not presentation tools, and the primary audience of a prototype needs to be the user.
Service design playbook | ontario.ca
Alpha is not about creating presentation tools for stakeholders
Service design playbook | ontario.ca
Alpha is not about building the solution
Service design playbook | ontario.ca
identify existing processes or policies that will need to change to support the service
Service design playbook | ontario.ca
Not all projects should go beyond discovery. It’s acceptable – and sometimes advisable – to end a project at this phase if that’s what the evidence supports.
Service design playbook | ontario.ca
Use research even if it doesn’t align with expected findings, to show how far a problem might extend.
Service design playbook | ontario.ca
It can be tempting to skip discovery and design a solution based on similar services or brainstorm potential options immediately. These are solution-oriented design processes, but service design is a user-oriented design process.
Service design playbook | ontario.ca
All that’s needed is a piece of paper, a pencil and some elbow grease.
Service design playbook | ontario.ca
With that in mind, all public servants – whether they work in digital, communications, policy or operations – are involved in service design.
Service design playbook | ontario.ca
Ontario’s Anti-Racism Act
Digital Service Standard | ontario.ca
Ontario’s Simpler, Faster, Better Services Act
Digital Service Standard | ontario.ca
integrate automated testing tools into the processes for development and maintenance
Digital Service Standard | ontario.ca
use a testing environment and context as close as possible to those that users will experience
Digital Service Standard | ontario.ca
design and test the service or product to work with users’ browsers and devices, including assistive devices
Digital Service Standard | ontario.ca
document how the service or product was built and how it will be maintained, including how this documentation will be kept up-to-date
Digital Service Standard | ontario.ca