Rocket Pro Navigate: Conducting Rapid Research to Uncover AI Use Cases for Mortgage Brokers
Rocket Pro Navigate: Conducting Rapid Research to Uncover AI Use Cases for Mortgage Brokers
Rocket Mortgage
|
UX Research Intern
May 2025 - August 2025
Timeline: 8 weeks
Role: UX Research Intern
Tools: dScout, Figma, Azure DevOps
Methods: User Interviews, Diary Study, Lean Agile
Disclaimer: Some details about this project may be limited due to a Non-Disclosure Agreement
Problem: Mortgage brokers use LLMs inconsistently and often outside Rocket’s ecosystem. Rocket needed to understand how brokers currently use AI tools—and whether a Rocket-branded LLM could genuinely assist their workflows.
Goal: Test a working prototype with brokers to learn if and how they would adopt a Rocket-built LLM, what tasks they would trust it with, and what improvements were needed before wider rollout.
Outcomes:
Ran iterative, rapid discovery research directly with a functioning LLM
Identified early usability issues and broker expectations before launch
Uncovered two high-value application concepts that were incorporated into the LLM
Insights, paired with Rocket Assist research, helped inform the broader strategy for chat-based AI tools
Reflection: Insights, paired with Rocket Assist research, helped inform the broader strategy for chat-based AI tools
Research Deep Dive
Relevant Products
Rocket Pro
Rocket Pro is a digital tool for third-party mortgage brokers and loan officers to manage and monitor loan statuses when their clients have loans being processed with Rocket Mortgages and other lenders.
They use it to track loans and to submit all the necessary documents to ensure a timely closing.
Rocket Pro is a digital tool for third-party mortgage brokers and loan officers to manage and monitor loan statuses when their clients have loans being processed with Rocket Mortgages and other lenders.
They use it to track loans and to submit all the necessary documents to ensure a timely closing.
Rocket Pro Navigate
Rocket Pro Navigate is a recently released LLM tool for mortgage brokers and loan officers that helps them with their everyday work tasks. It can do anything from analyzing documents and providing insights into next steps and discrepancies, to prioritizing and drafting communications for leads, and much more.
Rocket Pro Navigate is a recently released LLM tool for mortgage brokers and loan officers that helps them with their everyday work tasks. It can do anything from analyzing documents and providing insights into next steps and discrepancies, to prioritizing and drafting communications for leads, and much more.
Background
The Rocket Pro Navigator project is a prime example of a quick and scrappy product pilot. In my 12 weeks at Rocket, Pro Navigator was discussed in a meeting during week 3, I learned about it in a coffee chat during week 4 and asked to join the project, and by week 8, we had it up and running to test with users. The only hurdle we faced was: "Would they use it? How would they use it? What specific information or tasks would be most beneficial for mortgage brokers?" This is where the research I helped co-pilot came into play. There were a lot of pivots and changes to the research, design, development, and eventual testing. For me, it was an exercise in Lean Product Management.
The Rocket Pro Navigator project is a prime example of a quick and scrappy product pilot. In my 12 weeks at Rocket, Pro Navigator was discussed in a meeting during week 3, I learned about it in a coffee chat during week 4 and asked to join the project, and by week 8, we had it up and running to test with users. The only hurdle we faced was: "Would they use it? How would they use it? What specific information or tasks would be most beneficial for mortgage brokers?" This is where the research I helped co-pilot came into play. There were a lot of pivots and changes to the research, design, development, and eventual testing. For me, it was an exercise in Lean Product Management.
The Rocket Pro Navigator project is a prime example of a quick and scrappy product pilot. In my 12 weeks at Rocket, Pro Navigator was discussed in a meeting during week 3, I learned about it in a coffee chat during week 4 and asked to join the project, and by week 8, we had it up and running to test with users. The only hurdle we faced was: "Would they use it? How would they use it? What specific information or tasks would be most beneficial for mortgage brokers?" This is where the research I helped co-pilot came into play. There were a lot of pivots and changes to the research, design, development, and eventual testing. For me, it was an exercise in Lean Product Management.
Week 3
Week 3
Week 3
Inspiration Sparked
Inspiration Sparked
Inspiration Sparked
Week 4
Week 4
Week 4
Jumping In
Jumping In
Jumping In
Week 8
Week 8
Week 8
Launch & Learn
Launch & Learn
Launch & Learn
My involvement with this project also had its own unique challenges. On the standing Rocket Pro research team, the individual leading research for Pro Navigator was set to go out of the office for 3 weeks. Since the project became a reality so quickly, no one else had had the chance to jump in, except for me. After I spoke with the team member going on vacation and my leader, I volunteered to lead the project for research while my colleague was out of the office. With their vote of confidence and my teammates' proactiveness in ensuring all decisions made up until that point were well documented, I became the research point of contact for the duration of my internship.
My involvement with this project also had its own unique challenges. On the standing Rocket Pro research team, the individual leading research for Pro Navigator was set to go out of the office for 3 weeks. Since the project became a reality so quickly, no one else had had the chance to jump in, except for me. After I spoke with the team member going on vacation and my leader, I volunteered to lead the project for research while my colleague was out of the office. With their vote of confidence and my teammates' proactiveness in ensuring all decisions made up until that point were well documented, I became the research point of contact for the duration of my internship.
Research Approach
Research Goal
Research for Pro Navigator was unique in that it took a more assumption-centric approach, rather than an exploration-centric approach. This was largely because the product was mostly already built; it just needed tweaks to become external-facing. Thus, we decided to focus on giving it to the users and letting them freely experiment. We wanted to know whether they'd use it, what they would do with it, and what they wished they could do with it.
Research Goal
Research Goal
Learn about how mortgage brokers and loan officers feel about and would interact with an LLM built to cater to the needs of their roles.
Learn about how mortgage brokers and loan officers feel about and would interact with an LLM built to cater to the needs of their roles.
Learn about how mortgage brokers and loan officers feel about and would interact with an LLM built to cater to the needs of their roles.
Methods
Study Design
There were also many changes in the research design, up until the week before we began testing. This was largely because there were numerous hoops to jump through, including reviews from information security and legal departments. Thus, there were often adjustments with who our participants would be, how many there would be, and what they'd be able to do with the application.
There were also many changes in the research design, up until the week before we began testing. This was largely because there were numerous hoops to jump through, including reviews from information security and legal departments. Thus, there were often adjustments with who our participants would be, how many there would be, and what they'd be able to do with the application.
Initial Plan:
Timeline: 2 weeks
Participants: 10-15 mortgage brokers who are new to AI and 10-15 mortgage brokers with some experience with AI
Survey: Distributed at the beginning to learn about their experiences with AI tools
Diary Study: Participants will have access to Pro Navigator for 2 weeks. Every day, they will be asked to submit a summary of their usage. After the first week, we will interview them and provide a tutorial. After the second week of use, we will conduct final interviews.
Initial Plan:
Timeline: 2 weeks
Participants: 10-15 mortgage brokers who are new to AI and 10-15 mortgage brokers with some experience with AI
Survey: Distributed at the beginning to learn about their experiences with AI tools
Diary Study: Participants will have access to Pro Navigator for 2 weeks. Every day, they will be asked to submit a summary of their usage. After the first week, we will interview them and provide a tutorial. After the second week of use, we will conduct final interviews.
Pivot Plan:
Timeline: 1 weeks
Participants: 3-10 mortgage brokers
Survey: Distributed at the beginning to learn about their experiences with AI tools
Interviews: Participants will have access to Pro Navigator for 1 week, with an interview after a week of use to learn about their experience.
Pivot Plan:
Timeline: 1 weeks
Participants: 3-10 mortgage brokers
Survey: Distributed at the beginning to learn about their experiences with AI tools
Interviews: Participants will have access to Pro Navigator for 1 week, with an interview after a week of use to learn about their experience.
It is important to note that with the Pivot Plan that was taken for the study, we decided the best way to start getting insights from as many users as soon as possible was to start with a small group approved by the information security team, and then keep increasing the participant pool as they approved more participants. Thus, the first wave had 3 users, then 6 more.
We also decided to abandon the diary study due to the small participant pool and lack of incentives. We needed to ensure we would get insights from this research, and any risk of dissuading participants from completing the study had to be mitigated, even if the diary study would have been more informative.
It is important to note that with the Pivot Plan that was taken for the study, we decided the best way to start getting insights from as many users as soon as possible was to start with a small group approved by the information security team, and then keep increasing the participant pool as they approved more participants. Thus, the first wave had 3 users, then 6 more.
We also decided to abandon the diary study due to the small participant pool and lack of incentives. We needed to ensure we would get insights from this research, and any risk of dissuading participants from completing the study had to be mitigated, even if the diary study would have been more informative.
AI Experience Survey
As our product team onboarded research participants, we also had them distribute a survey link so we could establish their background with AI. These surveys were not anonymous, so we could connect their interview data to their responses. This would allow us to understand the bigger picture of how varied AI experiences and sentiments may affect a users' adoption of a new AI tool. If we find that users who have a lot of experience with AI and use it regularly report negative experiences with the tool, as well as the expected AI detractors, that tells us that we may need to go back to the drawing board.
As our product team onboarded research participants, we also had them distribute a survey link so we could establish their background with AI. These surveys were not anonymous, so we could connect their interview data to their responses. This would allow us to understand the bigger picture of how varied AI experiences and sentiments may affect a users' adoption of a new AI tool. If we find that users who have a lot of experience with AI and use it regularly report negative experiences with the tool, as well as the expected AI detractors, that tells us that we may need to go back to the drawing board.
Sample Questions:
Which AI tools have you used before?
1. ChatGpt, 2. Perplexity, 3. Claude, 4. DeepSeek,
5. Llama, 6. Copilot, 7. Other(Describe). 8. None.
Have you paid for a subscription to any of these tools?
1. ChatGpt, 2. Perplexity, 3. Claude, 4. DeepSeek,
5. Llama, 6. Copilot, 7. Other(Describe). 8. None.
On a scale from 1 - 5, where 1 = very negative and 5 = very positive, how would you rate your overall experience using AI tools?
About how long have you been using AI tools?
How often do you use AI tools in your personal life (i.e., outside of work?)
Which AI tools have you used before?
1. ChatGpt, 2. Perplexity, 3. Claude, 4. DeepSeek,
5. Llama, 6. Copilot, 7. Other(Describe). 8. None.
Have you paid for a subscription to any of these tools?
1. ChatGpt, 2. Perplexity, 3. Claude, 4. DeepSeek,
5. Llama, 6. Copilot, 7. Other(Describe). 8. None.
On a scale from 1 - 5, where 1 = very negative and 5 = very positive, how would you rate your overall experience using AI tools?
About how long have you been using AI tools?
How often do you use AI tools in your personal life (i.e., outside of work?)
1 Week Post-Use Interview
After the participants had a week with Pro Navigator, we met with them to get their thoughts on the tool. We wanted to see if and how they would interact with it without any direction from our team. By knowing this type of information, we could find out what was missing from the product, how well it performed, and if there was even any desire to use it in their current workflows.
Combined with the survey data, this information would be even more valuable, because we could understand more about their background with AI. From here, we could seek any patterns between previous experience and adoption of Pro Navigator.
After the participants had a week with Pro Navigator, we met with them to get their thoughts on the tool. We wanted to see if and how they would interact with it without any direction from our team. By knowing this type of information, we could find out what was missing from the product, how well it performed, and if there was even any desire to use it in their current workflows.
Combined with the survey data, this information would be even more valuable, because we could understand more about their background with AI. From here, we could seek any patterns between previous experience and adoption of Pro Navigator.
Sample Questions:
Before utilizing Pro Navigator, how have you used AI before?
What tools have you used?
What was your workload like this week? Was it typical, busier than usual, or slower?
How would you describe your overall experience using Pro Navigator?
How did you learn to use the Navigator Pro during the week?
What specific tasks or questions did you use Navigator Pro for this week?
Can you share your screen with me and walk me through your process?
Before utilizing Pro Navigator, how have you used AI before?
What tools have you used?
What was your workload like this week? Was it typical, busier than usual, or slower?
How would you describe your overall experience using Pro Navigator?
How did you learn to use the Navigator Pro during the week?
What specific tasks or questions did you use Navigator Pro for this week?
Can you share your screen with me and walk me through your process?
Participants
Gathering participants for our research was another aspect of the project that involved a lot of pivoting and changes. As I mentioned previously, there were initially meant to be up to 30 participants recruited by the sales team. Limitations in place by information security and legal then lowered this number to no more than 10.
At the end of the day, gathering data was the most important goal. Even though we were operating with a smaller batch of participants, we needed to make sure we could reliably gather input from them, on an even shorter time frame than before because of delays in the pilot launch.
Gathering participants for our research was another aspect of the project that involved a lot of pivoting and changes. As I mentioned previously, there were initially meant to be up to 30 participants recruited by the sales team. Limitations in place by information security and legal then lowered this number to no more than 10.
At the end of the day, gathering data was the most important goal. Even though we were operating with a smaller batch of participants, we needed to make sure we could reliably gather input from them, on an even shorter time frame than before because of delays in the pilot launch.
Outcomes
Study Participation Totals:
In total, we had 10 participants utilizing Rocket Pro Navigate over the course of the product pilot. We collected analytics from these users, including how often they were logging in, their prompting frequency, and any custom applications they made. We were able to gather insights into the previous AI usage of 3 participants, and gather qualitative feedback from 4 participants.
In total, we had 10 participants utilizing Rocket Pro Navigate over the course of the product pilot. We collected analytics from these users, including how often they were logging in, their prompting frequency, and any custom applications they made. We were able to gather insights into the previous AI usage of 3 participants, and gather qualitative feedback from 4 participants.
Survey Results
Interestingly, I was simultaneously interviewing mortgage brokers about AI tools for Rocket Pro Assist while also surveying them on their use of AI tools for Rocket Pro Navigate.
This allowed me to reference some of the more general AI usage data points for both projects. Thus, I had data on general AI usage patterns and sentiments from 6 mortgage broker interviews and 3 mortgage broker surveys.
Interestingly, I was simultaneously interviewing mortgage brokers about AI tools for Rocket Pro Assist while also surveying them on their use of AI tools for Rocket Pro Navigate.
This allowed me to reference some of the more general AI usage data points for both projects. Thus, I had data on general AI usage patterns and sentiments from 6 mortgage broker interviews and 3 mortgage broker surveys.
This is unsurprising considering how diverse this population is. We spoke to a 63 year-old broker from California who loved AI, a 56 year-old who hated it, a 29 year-old who loved it, and surveyed 3 who were all lukewarm on it.
This allowed us to understand the challenge of meeting a diverse population where they are in terms of AI adoption.
This is unsurprising considering how diverse this population is. We spoke to a 63 year-old broker from California who loved AI, a 56 year-old who hated it, a 29 year-old who loved it, and surveyed 3 who were all lukewarm on it.
This allowed us to understand the challenge of meeting a diverse population where they are in terms of AI adoption.
Across the board, no matter their initial sentiments or usage of AI, the top concerns were always accuracy and accountability. This was largely because they worried they would miss something AI had done incorrectly for them, or it wouldn't be able to do what they were asking it to do, then there wouldn't be a human to reach out to for help.
This allowed us to understand how previous experiences have shaped our users' perceptions of AI, particularly LLMs like ChatGPT or chatbots.
Across the board, no matter their initial sentiments or usage of AI, the top concerns were always accuracy and accountability. This was largely because they worried they would miss something AI had done incorrectly for them, or it wouldn't be able to do what they were asking it to do, then there wouldn't be a human to reach out to for help.
This allowed us to understand how previous experiences have shaped our users' perceptions of AI, particularly LLMs like ChatGPT or chatbots.
Feedback Gathered
At this point, we had to return to only relying on the data we gathered from this study's participants. We now knew that, in regard to this particular study, the 3 survey respondents had at least some AI experience, mostly with ChatGPT. We also knew they had neutral sentiments about AI, with some concerns for accuracy and accountability.
After 1 week of having access to Rocket Pro Navigate, we received the following major insights from 4 study participants. 1 of those who did not complete the survey, but was very enthusiastic and a frequent AI user.
At this point, we had to return to only relying on the data we gathered from this study's participants. We now knew that, in regard to this particular study, the 3 survey respondents had at least some AI experience, mostly with ChatGPT. We also knew they had neutral sentiments about AI, with some concerns for accuracy and accountability.
After 1 week of having access to Rocket Pro Navigate, we received the following major insights from 4 study participants. 1 of those who did not complete the survey, but was very enthusiastic and a frequent AI user.
In this current iteration, 3 out of the 4 participants mentioned that they felt this tool would be best suited for newer mortgage brokers
In this current iteration, 3 out of the 4 participants mentioned that they felt this tool would be best suited for newer mortgage brokers
This was largely because it could help answer questions about guidelines, craft emails, and assist with cold calling, not so much with specific time-intensive tasks.
This was largely because it could help answer questions about guidelines, craft emails, and assist with cold calling, not so much with specific time-intensive tasks.
All 4 of the brokers mentioned ways in which the product could have more specific use cases to the average mortgage broker
All 4 of the brokers mentioned ways in which the product could have more specific use cases to the average mortgage broker
Identifying paths forward for Navigator was a major goal of this research. By gathering commonalities in the suggestions, we could iterate for the next round of testing.
Identifying paths forward for Navigator was a major goal of this research. By gathering commonalities in the suggestions, we could iterate for the next round of testing.
Inaccuracies in response were reported by 2 participants. Another participant expressed concern about the company's privacy
Inaccuracies in response were reported by 2 participants. Another participant expressed concern about the company's privacy
This revealed to us that we needed to refine the knowledge base. We also realized we needed clearer data storage info and settings for companies to control.
This revealed to us that we needed to refine the knowledge base. We also realized we needed clearer data storage info and settings for companies to control.
Recommendations
In a meeting with the product manager, the product design leader, the product design manager for Pro Navigator, and my teammate that had returned from his PTO, we had a discussion about what the future for Pro Navigator could be.
Three major recommendations I had for Rocket Pro Navigate applied to all LLM-style chat products operating within the Rocket Pro space: We needed to find ways to meet brokers where they are in terms of AI experiences by providing more value with less effort on their part, and by potentially expanding into different user experiences. We also needed to consider unifying the chat experiences in the Pro space, if research revealed user confusion.
In a meeting with the product manager, the product design leader, the product design manager for Pro Navigator, and my teammate that had returned from his PTO, we had a discussion about what the future for Pro Navigator could be.
Three major recommendations I had for Rocket Pro Navigate applied to all LLM-style chat products operating within the Rocket Pro space: We needed to find ways to meet brokers where they are in terms of AI experiences by providing more value with less effort on their part, and by potentially expanding into different user experiences. We also needed to consider unifying the chat experiences in the Pro space, if research revealed user confusion.
