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
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