AI Assets

Assets for Building Successful Products with AI

Pick a Product

Pick one of these two products. It will be your product for the rest of session.

1. AI-Driven Loan Application Assistant

  • Goal: Build a prototype that streamlines loan applications by using AI to predict approval likelihood, identify missing documents, and provide proactive insights.

  • Best Suited For: Participants interested in automation, predictive analytics, risk assessment, and workflow optimization.

2. AI-Powered Savings App

  • Goal: Build a prototype savings app that leverages AI to personalize savings goals, recommend strategies for better financial habits, and offer tailored insights based on spending behavior.

  • Best Suited For: Participants interested in consumer-facing products, personalized recommendations, financial wellness, and engaging UI experiences.

Exercise 3

Download the CSV that corresponds with the product you selected.

AI-Driven Loan Application Assistant

https://drive.google.com/file/d/1fgzJAPTtPlu3fEymG3AzylZhuJ76G6Up/view?usp=drive_link

AI-Powered Savings App Data

https://drive.google.com/file/d/1gHBDVsmcmSgF4yiYcqSW1x9ziR5Qx_3x/view?usp=drive_link

Directions

  1. Create a new chat within your Claude project.

  2. Review the data file to see what type of data is included in it.

  3. Upload the file to your chat.

  4. Ask a question like “Show me which features correlate most strongly with retention?”

  5. Imagine you have to present progress of the app from the last month to the exec team. Ask it for a graph that you want to include in your exec presentation.

  6. Ask it a question that would give you ideas of what areas you should work on next.

  7. Share your prompts and output with your neighbor and discuss.

Exercise 4

Access the four user interviews associated with your product.

AI-Driven Loan Application Assistant User Interviews

https://drive.google.com/drive/folders/1xwWQIOJGVV_AMbA3QnubLem3edV-7IQ3?usp=drive_link

AI-Powered Savings App User Interviews

https://drive.google.com/drive/folders/1QgcSTieb_XbopDPW2sUIY1FGl6iXkQ3D?usp=drive_link

Directions

  1. Put the contents of each interview into Claude and follow the directions below.

  2. Prompt the LLM:

    • Ask the LLM “What should I build?”

    • Now instruct the LLM to forget what you just asked and what it just shared with you. Write a new prompt that has it break down what assumptions it validated, analyze the why behind what the interviewees said, and outline the areas we should dig deeper on.

    • Now ask it share some options we could explore based on one of the insights.

    • Discuss with the person next to you the impact of asking the LLM good questions

Exercise 5

Directions

  1. Open V0.

  2. Create a prompt for your prototype.

  3. Ask Claude or Perplexity for feedback on your prompt.

  4. Adjust your prototype as needed.

  5. Explain to your neighbor the direction you chose for your prototype, what the AI got right, and the things it didn’t get right or took some adjusting.

  6. Take turns giving feedback on the prototype from design choices to asking questions about where some of the data might come from.

  7. Share with your neighbor one assumption you want to test with customers with this prototype

Exercise 6

Directions

Put the contents of each usability interview into Claude and follow the directions below:

  • Prompt the LLM:

    • Ask Claude to identify the top 3 usability barriers encountered during the prototype testing

    • Instruct Claude to analyze the root causes behind these friction points

    • Ask Claude to highlight any unexpected user behaviors or comments that might reveal deeper product opportunities

  • Discuss with your partner how these insights could transform your prototype and accelerate your path to product-market fit

Exercise 7

Scenario: Your developer says they need detailed designs for login error states for the app. Your designer doesn’t have bandwidth to take time on this.

Directions

  • Write out how you would approach this situation, including any tools or people you would involve.

  • Create the needed error states.

  • Discuss with your neighbor you approach, show them the error states, and exchange feedback.

Exercise 8

Loan AI PRD

https://docs.google.com/document/d/1orNL4UOJ7qx4y2iDdbHEalZnfGXrBhLYhCCb_XZsIM4/edit?usp=drive_link

Savings AI PRD

https://docs.google.com/document/d/1vIsSPjluwmdovd6vnTMRzIbf4IEcZI-BHVjpAZuYYDE/edit?usp=drive_link

Directions

Review the AI PRD for the product you chose.

  • Read the PRD.

  • What is your initial thought when reading it?

  • When you look closer, what are things that need improvement.

Exercise 9

Directions

Take a screenshot of at least one feature from your app in V0.

Other Links

Generative Agent Simulations of 1,000 People

https://github.com/joonspk-research/genagents