Inspiration
Planning rules for small home projects are hard to navigate for non-experts. We focused on a real local pain point: Cambridge homeowners trying to answer “Can I do this work, and what should I do next?” without jumping across multiple policy websites and maps.
What it does
Camwall is a planning triage assistant for common residential changes like walls, sheds, extensions, rooflights, and hard surfaces. Users pick one of five Cambridge demo properties, answer structured questions, and get a deterministic verdict with clear reasons, assumptions, sources, and next steps. It also exports a downloadable report in Markdown and PDF.
How we built it
We built a deterministic rules engine in Python with separate modules per proposal type, then layered an explanation/report composer on top. The frontend is a multi-step web flow (property -> context -> proposal -> questions -> review -> report/compare). Data is hardcoded for reliability in demo conditions, and export is markdown-first so the same report content powers both .md and .pdf.
Challenges we ran into
The main challenge was balancing realism with hackathon reliability. Planning logic is nuanced, so we narrowed scope to high-signal rules and made escalation behaviour explicit for conservation/listed cases. We also avoided fragile external dependencies to keep demos stable.
Accomplishments that we're proud of
We shipped a full end-to-end product flow, not just a rules demo. The system clearly separates deterministic decisions from generated explanation text, includes transparent citations, and provides practical next-step routing. We also added scenario comparison and report export, which made the demo feel production-oriented.
What we learned
Trust in AI tools comes from constraints, transparency, and clear boundaries. Deterministic logic plus structured inputs is much easier to explain and test than open-ended prompting alone. We also learned that curated datasets can make early-stage products much more reliable and persuasive.
What's next for Camwal
Replace hardcoded properties with live Cambridge address search and real planning-constraint lookups. Expand proposal coverage (porches, dormers, more roof/extension variants) and improve local policy depth. Add saved reports, user accounts, and optional professional handoff to planners/architects for complex cases.
Built With
- python?
Log in or sign up for Devpost to join the conversation.