A Small Experiment in Mobile AI and Valuation
CapRate Vision Lite is a prototype that explores how mobile AI can support reasoning about risk, income, and value using something as simple as a building photo. Instead of starting from a formula or a spreadsheet, the app begins with a real or everyday image and asks an AI model to construct a teaching example around it.
When you open the app, you upload a photo of a building that could plausibly earn rent—shops on a main street, a small apartment building, or a mixed-use property. You can optionally supply a rough annual income and a city or location. The system then sends this information to a multimodal AI model (Gemini), which returns:
- a brief description of what it sees,
- a hypothetical cap-rate range,
- an illustrative value range (when income is provided), and
- a short explanation of why that range might make sense, followed by a reflection question.
The intention is not to produce a market-accurate valuation, but to create concise, critique-able narratives that sit halfway between formula and practice. By tying each scenario to a concrete image and a visible reasoning trail, learners can experiment with different properties, compare cases, and discuss where they agree or disagree with the AI’s story. The reflection prompts at the end of each run are designed to nudge users toward making their own judgments explicit—an important step in developing professional reasoning in any domain.
CapRate Vision Lite is built as a single-page, mobile-first web app using Firebase, Gemini, and a small amount of React. It is deliberately constrained to one narrow concept (cap rates) and one simple interaction pattern (image → narrative), but the same pattern could be extended to other topics where visual inspection and risk trade-offs matter.
You are invited to open the prototype in a browser, upload any suitable building photo, and see how the AI frames its teaching example.
Launch the app: CapRate Vision Lite
Love the direction you went in! Although my personal understanding of business and evaluation is not the strongest, I can definitely see the practically it of this tool!
From an education perspective, what really stood out to me is how CapRate Vision Lite could totally change the way students learn business concepts. I know how hard it can be for learners to connect numbers on a page to what’s actually happening out in the world. Students especially struggle to see how location can affect pricing. But the idea of pointing your phone at a building and instantly getting a cap rate story you can pick apart makes it feel very practical. It’s the kind of hands-on learning and application that promotes understanding because you’re seeing the logic in context, not just working through a formula.
I also love that the AI doesn’t pretend to be right, but rather it gives a starting point and lets the learner push back, which feels closer to how decisions get made in actual business environments. There’s always room for debate and interpretation. A tool like this could help students build that confidence to question things and justify their thinking. And since it’s mobile, it fits the way people in business actually work, grabbing little moments to learn while they’re on the go.
The CapRate Vision Lite is a genuinely engaging approach to teaching complex valuation concepts, it demonstrates how learning designer can also use case studies way that supported by AI to guide learning. It translates the abstract/ challenging concepts or math formulas into a real-world, personalized context. It perfectly aligns with your stated justification, “CapRate Vision Lite can be read as an early, tightly scoped instance of multimodal agency in education: an AI ‘looks’ at a scene, constructs a narrative about risk and return, and then hands the reasoning back to the learner as something to critique, extend, or reject.”
It represents a new frontier for multimodal AI in education, transforming passive consumption into active critique. It would become even more powerful if the reflection questions were made interactive, ie. the learner could be prompted to type in their response and interact with the AI assistant to learn more.
Thanks again for your wonderful presentation.
I really enjoyed exploring CapRate Vision Lite. What stood out to me is how it makes an abstract idea feel concrete by starting with an image instead of a formula. Like others mentioned I agree—this isn’t about getting a perfect valuation. The real value is in the narrative the AI creates and how learners can question it. That “image → explanation → reflection” pattern feels powerful, especially for students who need a bridge between theory and real-world reasoning.
I also liked the point about how adaptable this could be across subjects. Even in areas far from real estate—like trades or design, and even more abstract fields—the idea of grounding a concept in something visual and familiar gives students a clearer starting point. It supports the shift we’re seeing where learners separate learning, productivity, and enjoyment, and want tools that help them move between these modes more naturally.
This prototype feels like a small example of how mobile AI can help students build judgment, not just receive answers. That’s a direction I’d love to see grow in more educational apps.
This is an interesting prototype that could have an interesting role in the classroom. Although I would have trust issues around the accuracy I could still see it having significant value as a teaching/practice tool. If tied to a custom LLM the value and potential could be significant — across various contexts. This is a really interesting application.
Thanks for sharing all these great thoughts on CapRate Vision Lite! I love how you pointed out that core idea of feeding AI info, letting it build a narrative, and handing it back for us to tweak or push back on—that really clicked with your LifeOS app too. It’s cool to see that shared thread even across totally different topics.
You’re spot on about making abstract stuff like cap rates feel real and visual—instead of drowning in formulas, students get a concrete starting point to build from. That makes total sense for any tough subject, and it’s awesome you had to look it up yourself – shows exactly why tools like this are needed!
Hi Shawn,
This was a very interesting way of combining AI vision models and contextual pedagogy to support learning. As someone who has used Apple Intelligence to figure out whether the plant my dog might have eaten is toxic, the idea that this can be effectively integrated into the classroom is intriguing. The ability for AI to modify how it communicates (i.e., for a student in elementary vs. high school) would help clarify concepts to meet the needs of the learner. In terms of context, I agree with your thought process regarding the importance of real-life connections and in prompting reflection to further develop critical thinking. I envision CapRate Vision Lite in the trade classes (i.e, mechanics, culinary, etc.). What role would it play in more abstract classes (i.e., music)?
Hi Shawn,
I really enjoyed viewing your prototype! While I don’t have a lot of familiarity with this area, I could see the potential for this in educational and real-world contexts. This would be very engaging for students to work with to get a sense of how properties can be analyzed. Overall, great concept!
This is a really fascinating proof of concept for a very useful app. While I’m slightly uncertain about its integration into a typical classroom, I do see the educational value from this app in certain settings. The app itself is very clearly laid-out and uses strong design principles that would interest users. I also appreciate that you have included examples for users who do not have a photo to upload. The variables you provide for date entry with the photo is an important part of making the prompt successful.
Hi Shawn
I really enjoyed reading about your CapRate Vision Lite prototype. What stood out to me was how you take a concept as abstract as capitalization rates and anchor it in something students can actually see and make sense of. Admittedly I had to look up what a cap rate was before diving in, and that only reinforced how much a tool like this can be helpful for newbies. It gives learners a starting point that feels concrete instead of overwhelming, and it creates space to build intuition before they’re expected to handle formulas or market data. It also reminded me how disorienting many technical subjects can be when students are asked to master theory without any meaningful context. Tools that connect abstract ideas to visuals, plain language, and real-world cues give students a way in (regardless of discipline) and this prototype really speaks to that broader need for accessible, grounded learning.
Hey there sdavis18.
While I was surprised to see a submission like this that wasn’t obviously tied to formal education, I was quick to realize that the heart of the rationale behind your proposed technology was similar to mine. The exact sentence that solidified this for me was in your forecast on mobile intelligence; “CapRate Vision Lite can be read as an early, tightly scoped instance of multimodal agency in education: an AI “looks” at a scene, constructs a narrative about risk and return, and then hands the reasoning back to the learner as something to critique, extend, or reject.” While the purpose behind my LifeOS app (a narrative continuity app for young learners to develop meaning in their experiences) seems vastly different, the underlying theory is the same. Feed an AI a curation of information, have it construct a narrative for you, and hand it back to you for you to reflect on, agree upon, or reject.
I think you did a great job researching and explaining the rationale behind your product and how it fits with the current state of the art in mobile intelligence.
I’m curious to know why you chose this use case to show your understanding of the forefront of mobile intelligence. Do you work in the real estate sector? If so, or regardless really, what significance does the future of mobile intelligence mean to you and/or your profession?
Cheers,
Jake