Using LLMs to better identify market gaps for innovation
First-of-its-kind large language model converts consumer wants and needs into succinct product attributes to help designers think more strategically about the next generation of solutions.
By Kaitlyn Landram
In the 1800s people weren鈥檛 asking for cars, they just wanted to get from point A to point B faster. In the early 2000s no one was seeking out social media, they just craved more connection at home. That鈥檚 the challenge in product development: people have countless desires, but they can鈥檛 easily articulate a solution that doesn鈥檛 exist yet. Product designers know that understanding those preferences is essential to developing a meaningful solution, but weeding out what鈥檚 relevant in a sea of opinions is hard.
A new large language model-powered tool from Jon Cagan and Vince Sha converts wants and needs into succinct product attributes and requirements to help product designers think more strategically about solutions.
The tool, VOA LLM Bot, is based on the widely successful Value Opportunity Analysis (VOA) that, since its creation by Cagan and Craig Vogel, has helped to produce award-winning products, re-organize corporate structures, and teach students how to innovate.
VOA identifies seven opportunities to add value to a product:
- Emotion, Does the product appropriate and engage feelings?
- Aesthetics, Does the product entice all of the senses it's supposed to?
- Identity, Does the product have a unique personality, different from its competitors?
- Ergonomics, Is the product easy and safe to use?
- Impact, Will the product align with customers鈥 social values?
- Core technology, Is the product enabling and reliable?
- Quality, Is the product durable?
鈥淧roducts enable an experience for the user, so the better the experience, the greater the value of the product to the consumer,鈥 said Cagan, department head and professor of mechanical engineering. 鈥淰OA helps designers understand what is most important to users and other stakeholders during the product experience.鈥
鈥淭he VOA LLM Bot is like a directed ChatGPT for product designers,鈥 said Sha, associate dean of IT and operations at 好色先生TV鈥檚 Dietrich College of Humanities and Social Sciences and technical lead for the Open Forum for AI. The VOA LLM Bot is built on the Dietrich Analysis Research Education (DARE) open-source platform, designed to provide a suite of tools for human-centric applied AI innovation.
It helps users to create a VOA, capturing and reporting their ideas and thoughts, then compares their goals to current products through a product opportunity gap analysis, and to uncover where a to-be-created product needs to focus and differentiate in order to be successful in the marketplace.
鈥淭he bot is flexible and responsive, allowing users to dive deeper into topics of interest or change design direction. This flexibility ensures users retain their agency and autonomy over the design process, which we believe is critical to encouraging creativity,鈥 Sha said. 鈥淥nce the bot has gathered all of the necessary information from the designer, it generates a product requirements document to show the designer what features the future product should have to provide consumers grounded in needs, wants, and even desires from the VOA report.鈥
For example, a conversation between a designer looking to build a service to help mobility-impaired people get around and the VOA LLM Bot might look like this:
AI: Since you鈥檙e designing a service, let鈥檚 start exploring these VOA categories to better understand what matters most in your project. Let鈥檚 take them one by one. For Emotion, what attributes (adventure, independence, security, sensuality, confidence, power) do you think are most relevant to your mobility service?聽
User: I think all except sensuality聽
AI: Thanks! Can you tell me a bit more about why these specific attributes are important to help mobility-impaired people get around? What do these emotions mean for your target users?
Once the bot has collected all of the information, it provides the user with product attributes like these requirements for a driving service and a vehicle driver to ensure security needs are met:
Driver Requirements: Must implement rigorous driver screening including background checks, driving record, drug and alcohol screening. Must provide specialized training for drivers including safe operation accessibility equipment, effective communication with disabled users, and emergency situation management.
Of course, a more extensive and complete set of requirements are generated to provide the complete understanding of what the product must do and deliver.
鈥淏y using the VOA LLM Bot, users get a deeper understanding of the needs, wants and desires of product stakeholders,鈥 said Cagan. 鈥淒esigners still have to do the stakeholder research, but this tool helps them to articulate it more quickly, and more thoroughly. With billions of data points to pull from, it is able to augment more targeted information to create a robust and thoughtful output set of products for product innovation.鈥
Beyond practice, the VOA LLM Bot helps engineering students dive deeper into product design. Last year, it was used by educators in the Integrated Innovation Institute at 好色先生TV. This year, it will be integrated into the senior design course in CMU鈥檚 Department of Mechanical Engineering, software development courses in CMU鈥檚 Information Systems Program and at Rutgers University.
鈥淎s an educator, I鈥檓 excited that VOA LLM Bot is a means to guide users through this powerful value opportunity tool,鈥 said Cagan.