IBM
UX Researcher
Building a genAI assistant to accelerate business activities and streamline productivity.
Long Term Project
SUMMARY
1
Background
2
The Product
3
Problem Evaluation
4
Spotlight Study
5
Outcomes
The product
AskIBM is a generative AI assistant designed to productivity by generating emails, blog posts, document summaries and translations.
Key responsibilities
Led UX research to define AI impact across roles, developing a framework to measure productivity and assess impact, showcased in the blog post case study below.
What made this unique
Working on the cutting edge of new genAI advancements and inventing a new research method to match this new technology.
AskIBM: CIO Case Study
AskIBM. The AI-powered digital assistant that’s changing the way IBM works
Empowering IBM employees with a reinvigorated intranet
The IBM CIO organization modernizes the w3 homepage with generative AI.
Senior VP Joanne Wright discusses gen AI impact
180,000 IBMers get hands-on experience using IBM watsonx to find new ways to accelerate our work.
THE PEOPLE (the best part)
AskIBM UX Team (Alphabetical)
Andy Yu
Anna Nguyen
Annette Tassone
Eugene To
Gabby Hoefer
Jason Telner
Julia Lauer
Kathleen O’Brien
Nimsi Guerrero
Patrick Clough
Phyllis Mancino
RJ Gervacio
UX Designer
UX Designer
Design Principal
Visual Designer
UX Researcher
UX Researcher
AskIBM Design Lead
Visual Designer
UX Research & Design
Senior Designer
UX Researcher
UX Designer
Stakeholders
Juan Encalada
Laura Rodriguez
Niloofar Salahi
Jay LaPlante
Jason Horwitz
Marc Wisniewski
Jill Fourie
Matt Lyteson
w3 Design Lead
Product Strategist
Product Owner, AskIBM
Product Lead, AI & Automation
Head of w3 Engineering
Head of Design, AI & Automation
Vice President, AI & Automation
IBM CIO
THE PRODUCT
AskIBM: A virtual assistant bringing generative AI to the workplace
Building an AI-first digital assistant that could automate day-to-day business tasks such as drafting an email, translating a document or searching for information in order to enable all IBMers to focus on more strategic work
60 days
The AskIBM foundation was built in 60 days with watsonx.ai
30k+ documents
Over 30,000 documents were ingested into AskIBM
Timeline of AskIBM's launch journey, highlighting key milestones and achievements.
AskIBM lives on the w3 intranet
AskIBM is an internal generative AI chatbot to help answer general questions and productivity workflows.

Living at the top of the w3 intranet, AskIBM serves as the primary entry point for IBMers to get started using generative AI.
Timeline of AskIBM's launch journey, highlighting key milestones and achievements.
AskIBM architecture: built with watsonx.ai
The AskIBM foundation was built with watsonx.ai and IBM Granite LLMs. 30k+ strategic internal documents were ingested, enabling IBMers to automate normal business activities, such as drafting a client email, creating a product brief or quickly summarizing a long document using gen AI.
Timeline of AskIBM's launch journey, highlighting key milestones and achievements.
Timeline of AskIBM's launch journey, highlighting key milestones and achievements.
PROBLEM EVALUATION
Main design challenge
How might we design a genAI use case to help employees write blog posts?
Why blog posts?

3/4 of IBMers write blog posts quarterly. Blog posts communicate new releases, leadership updates and upcoming events. This is time consuming, relatively low-risk, and relies heavily on text-based outputs, making it an excellent use case for AskIBM to automate.
Key questions

1. What features do our users need?

2. What is the business value add for a blog post workflow in AskIBM?

3. What features are available on the market today?
USER RESEARCH SPOTLIGHT: USER INTERVIEWS
Timeline of AskIBM's launch journey, highlighting key milestones and achievements.
Qualitative
Exploratory user interviews into feature directions for AI-infused blog posts
Since this is a brand new feature, I built an interactive mural activity to help participants brainstorm solutions that could be possible with genAI and how that may solve their painpoints today.
Purpose
Identify painpoints for new generative AI workflow in AskIBM in order to streamline adoption and drive development decisions.

Approach
Completed 1:1 user interviews with applicants with 5 participants in Spring 2024.
  • 30-minute virtual sessions
  • Interactive mural activity
  • 5 blog-post writers
Mural illustrating users' blog post writing process and how Gen AI could enhance productivity and streamline writing.
Series of murals representing each participant's blog post writing process, with anonymized data, illustrating how Gen AI could improve productivity and streamline writing.

*Numbers and specific details have been fictionalized or removed due to confidentiality.

FINDINGS & OUTCOMES
The why: business value for AskIBM blog posts.
Blog posts is a promising use case for adding business value with AI. Working with the data science team, I brought my findings with numbers they collected in the Workplace Impact Survey.
  • Ideal level of difficulty: This a greater starter use case for AskIBM as only xx% of IBMers found writing blog posts to somewhat difficult or greater (n=2000+ from Workplace Impact Survey).

  • One participant already uses AskIBM for blogs: "With AskIBM, instead of taking me 2 months, it took me a week."
Diagram illustrating the business value of AI-related blog posts, with redacted data for IBM confidentiality.
The what: A roadmap with the top areas AskIBM’s genAI could impact blog post writing.
1. Getting started: Helping users get started by asking for specific inputs they already know was highlighted by xx% of participants, who found the process both time-consuming and highly important, with the first draft being the #1 most helpful feature.
2. Brainstorming title ideas: Empowering users to generate title options based on the generated blog content was highlighted by xx% of participants as one of the top five most useful challenges for AI to solve.
3. Editing and refining: Another top challenge is maintaining consistent voice and tone, which is especially important for IBMers who write blog posts on behalf of another author.
Diagram illustrating the business value of AI-related blog posts, with redacted data for IBM confidentiality.
The market: I also provided an analysis of top products on the market.
This analysis is IBM-confidential and cannot be shared publicly. I first created a user journey map for writing blog posts, then broke the market features down by each step in the journey.

This helped us to identify gaps in our current offerings and better understand the role that generative AI is playing in the blog post space today.
Data-redacted analysis showing a user journey map for writing blog posts, with market features broken down by each step in the journey. Highlights gaps in current offerings and the role of generative AI in the blog post space.