Rocket Lab Product Group Initiative

Role
User Experience Design Lead

Timeline
March 2023 – Now

Team Members
Program Manager
Dev Lead
Backend Dev
Frontend Dev

Initiative Overview

Our team is currently responsible for managing conversation automations and generative A.I. processes, which facilitate the connection between customers and advisors. Building upon the technology and expertise we have diligently cultivated, we aspire to establish a Product Group with the aim of broadening our client base and expanding our services.

Our present organizational structure and team function in a manner akin to an agency, wherein we receive requests from various business units within Apple and fulfill these requests in accordance with their specified formats. Nevertheless, with the introduction of the Rocket Lab Product Group Initiative, our aim is to transition into a product group paradigm. Under this new approach, our primary objective will be to conceptualize, develop, and refine products, actively seeking out business units within Apple that may find value in our offerings.

My Contributions

I have been assigned to this initiative in the capacity of the User Experience Design Lead. As part of our efforts, we have undertaken comprehensive exploratory user research, identifying distinct user personas and discerning their specific needs. Within the framework of this product group, our objective is to craft a unified product that has the potential to significantly enhance users' workflow and address their requirements effectively.

Who Are We?
Through the Rocket Lab Team initiative, we plan to showcase our expertise and create POCs, inviting external teams to collaborate with us.

Why?
Our team currently handles ad-hoc requests, leading to inefficiencies. With Rocket Lab, we're streamlining processes and defining a structured approach to product development.

Problem 1

Timeline
3 weeks

Main User Pain Point
"To create chat demos or to play around with LLMs, we need the team to make a fake bot or LLMs system, which takes over 2 months, sometimes delaying our business connections.”

Problem Statement
How can we develop a platform for pre-launch testing of conversation flows to minimize errors and eliminate redundant efforts in our production processes?

Exploratory Research

Participants
2 Product Managers
2 New Hires
2 Developer

Method
30 minute video call

Assumptions:

  1. There are teams that want to set up automated or not automated conversational experiences for their customers

  2. People that want to set up conversational experiences set it up by interface (if I want to automate an experience I already know if it should be messages vs web vs in app

  3. Our team is capable technologically of any flow proposed in this prototyping tool.

  4. Test chat in current existing product is insufficient for viewing bot dialogue flows

  5. Sharing a demo is more compelling in a realistically simulated environment than in a simple Keynote deck, Sketch file, or other static display

  6. Using this new product to convey a concept to business will be much quicker than using keynote, sketch, etc.

Research Questions:

  1. What has your process been to learn about Conversational Engineering products?

  2. What is your current process to create a demo of a chat experience? For example, if you wanted to create a Home Depot bot, how would you go about it?

  3. Can you give me an example of a feature or functionality you've developed recently (or are currently developing) that had a UI component to it? How did you visualize the UI as you were developing?

  4. How would having a tool like this have an impact on your process?

  5. If you had a magic wand, how would you change your current process?

User Archetypes

I want a tool that would allow me to get very hands on with in an experimental environment

Currently I have to add a prompt to the script and see the output in my terminal - a very brute force way

New Hires

Goals
See and feel what it’s like to converse with different bots with ML Models
Understand the components of a chat dialogue
Compare the response content from different ML Models

Tasks
Chat with various bots with ML models
Experiment with the different components of a bot dialogue

Wishes
It was easier to user our current tool in an experimental way not only for product purposes
Each ML model provided a web UI to interact with

Product Managers

Every Product Manager’s project is expanding to a lot of new business partners

Often new partners want to explore what is possible without committing to anything

Goals
Create lightweight proof of concept for business partners
Validate new features and experiences during pilot period

Tasks
Configure hardcoded bot flows for business partner use cases
Save and share experiences with business partners and QA

Wishes
A better way to show the end customer experience
More validation of ML models before the end customer experience goes live

Developer

We want to just work fast

Our toolset doesn’t lend itself to rapid collaboration [between developer and designers]

Goals
Rapidly prototype new features without cumbersome testing and deployment processes
Test the behavior of new bot dialogue flows
Collaborate with designers on UI for new features

Tasks
Use this new tool as a flexible frontend to surface new features
Interact with various bot dialogue flows
Save a snapshot of new features in this new tool frontend and share with designers

Wishes
Want a tool that enables rapid collaboration between developers and designers
Designers and other non-programmers could make changes in the UI and have those reflected in the code

Brainstorming

Final Iteration

Problem 2

Timeline
2 weeks

Exploratory Research

Generative AI/ML Engineer

Assumptions
People will provide accurate feedback to make the ML models better
This tool will decrease the time for generative AI engineer in making the models better
This feedback tool will provide accurate data to improve our response scripts

Problem Statement
How can we design a platform incorporating generative AI technology to effectively leverage our existing resources for providing continuous feedback and model retraining? This is aimed at consistently enhancing our generative assistant to ensure customers consistently receive exceptional experiences.

Needs
A way receive feedback on the ML models to retrain the models
Quickly prototype a conversation demo to get showcase what our team can do
Customize and view the UI of a chat before putting all money and efforts to production

Wants
Ability to share customized flow to other Product Managers to use
Seamlessly convert these into the production tool we use instead having to manually redo these chat flows in this prototyping tool

Individual Message Feedback Flow

Chat Experience Feedback Flow

Feedback Visualizer Flow