landing page of stadthelden-muenchen.de
matching page of stadthelden-muenchen.de

Inspiration

One of our team members had a great idea for a social project but hit a wall when trying to fund it. We realized that charitable foundations are surprisingly hard to find. Their information is often scattered, outdated, or buried in paperwork.

We saw a clear gap. There are resources available, but the bridge between "Idea" and "Funding" is broken. Since we had the right skills (AI, scraping, and full-stack development), we decided to build that bridge ourselves.

What it does

Stadthelden acts as a digital map and search engine for charitable foundations.

AI Discovery: We use AI and web crawlers to collect, classify, and normalize open foundation data.
Bureaucracy Hacking: We collaborate with Kulturator | Gute-Dinge-Stiftung to enable private individuals to receive project funding. This is usually only possible for established organizations.
The Antrags-Agent: A guided flow that helps users prepare the specific documents they need to apply for funding.

How we built it

We built a functional MVP in 36 hours by combining three core components:

The Scraper: A pipeline that visits foundation websites and turns messy HTML into semi-structured Markdown to get their mission, history, and requirements.
The Matcher: An algorithm that takes a user's project idea and ranks the best-fitting foundations.
The Agent: An interactive UI that assists the user in generating the required documents for their Förderantrag.

The Tech Stack:

Frontend: Next.js (Deployed on Vercel)
Backend: Python & FastAPI
Infrastructure: Self-hosted EC2 instance managing containers via Dokploy
AI/Data: crawl4ai to parse web content into Markdown for the LLM and Gemini to condense it down to useful information

Challenges we ran into

Data Chaos: Every foundation website is different. Defining a common data schema and extracting nested information from subpages was the hardest part of the backend work.
The Matching Logic: Tuning the algorithm was tricky. We had to ensure users don't just get any foundation, but the right one.
Going Live: We didn't want just a mockup. We struggled (and succeeded) to actually deploy the solution to a live domain during the hackathon.

Accomplishments that we're proud of

It is live: You can actually visit stadthelden-muenchen.de right now.
Full Pipeline: We successfully connected a Next.js frontend to a Python backend on a custom VPS within the time limit.
Real Value: We aren't just showing dummy data. We built a system that actually processes real foundation information.

What we learned

Dokploy is powerful: It is like an open-source Vercel that allows easy deployment on any VPS. This gave us full control over our backend.
LLMs prefer Markdown: We learned that using crawl4ai to convert webpages to Markdown is significantly more efficient for LLM processing than raw HTML.

Empowering the Community

We built this specifically with our home city in mind. Stadthelden isn't just software; it is social infrastructure for Munich. We are fundamentally changing how social engagement is funded and executed.

Relieving Administration: By empowering citizens to solve local problems independently, we reduce the direct burden on city administration while simultaneously increasing the quality of life for residents.
Combating Anonymity: In rapidly developing districts like Neufreimann, social isolation is a real risk. We foster a culture of "doing." We replace anonymity with active community bonds by making it easy for neighbors to fund and launch local initiatives.
Flipping the Script: Most platforms focus only on the beneficiaries. We focus on the change-makers. We empower the helpers, support the supporters, and enable the people who actually drive change.
Smart for Citizens: We turn days of frustrating bureaucratic research into minutes of clarity. The bridge between a "good idea" and "funded project" is no longer a maze. Now it's a straight line.
Helpful for Foundations: Foundations are often overwhelmed by irrelevant inquiries. Our matching algorithms ensure they receive high-quality, pre-vetted and highly relevant project proposals.

What's next for Stadthelden

Legal Finalization: We need to close the loop with Kulturator to fully operationalize the model for individuals and fully involve their offer into our workflow.
Expansion: We want to move beyond Munich to map the funding landscape for the entire DACH region.
Refinement: We plan to improve the "Antrags-Agent" so it can auto-fill even more specific forms.