Sadaqah Garden

Inspiration

The Prophet Muhammad ﷺ said that if a Muslim plants a tree and a bird, person, or animal eats from it, it is counted as charity for him (Sahih al Bukhari 2320). I kept thinking about that teaching while reading the news. The AI boom is reshaping the land itself. Large data centers can consume millions of gallons of water a day for cooling, often in regions already facing water stress, and their rapid expansion has converted farmland and pushed against communities that were there first. At the same time, pollinators are in trouble, with beekeepers in the United States reporting some of the worst colony losses on record in recent years, and stories of unexplained livestock deaths keep surfacing. The same technology I am building with carries real environmental costs, and I did not want to ignore that. I wanted to use AI sparingly and intentionally, to move people back toward the soil instead of deeper into screens. My faith tells me why this matters at the deepest level. The Quran teaches that Allah created humans from clay, and says: from the earth We created you, into it We will return you, and from it We will bring you out once more (Surah Ta Ha 20:55). We are made of this land, and it will be our home when we return to Allah. Reconnecting with it is not a hobby. It is remembering what we are. Sadaqah Garden closes that distance with one act: planting.

What it does

Food pantries serve nearly 48 million Americans, and the first thing they run out of is fresh vegetables, because people donate cans, not produce. Meanwhile, two thirds of America's 2,800 masjids already run food programs, and most sit on land that grows nothing but grass. Sadaqah Garden is the software that turns masjid lawns into pantry gardens. The unit is simple: one 4 by 8 raised bed grows about 60 pounds of tomatoes, peppers, cucumbers, and zucchini in a season, roughly 50 meals with fresh vegetables. Five beds fit on a typical masjid lawn, which means about 300 pounds a year for neighbors in need. The app runs the whole chain. Volunteers from the congregation sign up and choose roles: grow, water, harvest, or drive. Bustan, the AI garden coordinator powered by the Claude API, knows what is planted in every bed and generates the weekly task schedule, diagnoses sick plants, and advises on food safety for donations. When a harvest is logged, the app matches it to a nearby pantry by intake hours and requested produce within a 48 hour freshness window, a volunteer claims the delivery run like claiming a task, and the pantry confirms receipt by weight. Only verified pounds count, and every 5 verified pounds grows a new leaf on the masjid's Barakah Tree, an animated picture of the community's ongoing charity that its own donors can see.

How we built it

With the help of Claude, I built Sadaqah Garden as a React web app designed for mobile. The interface uses a natural palette informed by color research: deep forest green for trust and growth, warm cream for readability, and harvest gold reserved for action, since warm accents with high contrast draw attention and are associated with generosity. The app has four parts: a coordinator dashboard with live impact numbers and the Barakah Tree, a Beds view showing each bed's crops, expected yield, and the weekly tasks Bustan generates, the Bustan chat itself, and a Deliver flow that walks the four steps from logging a harvest to pantry verification. The Barakah Tree is an animated SVG I built by hand that grows leaves from verified donation data in a repeatable way, so the same history always renders the same tree. Bustan makes live calls to Anthropic's Claude API with a custom system prompt that gives it the garden's full context, keeps it focused on practical regional advice, and forbids it from inventing religious sources. Bed tracking, pantry matching, and the delivery flow run on structured demo data modeled on real Northern Virginia organizations.

Challenges we ran into

My biggest challenge was focus. The first version tried to serve every gardener with every feature, and judge feedback pushed me to narrow it to one quantifiable wedge: masjids growing for pantries, one bed at a time. Cutting features I loved, like individual sponsorships, made the product stronger. My second challenge was tone: making an app that is authentically Islamic without feeling preachy. I iterated on Bustan's system prompt so it connects gardening to tradition only where natural, and never invents hadith. On the technical side, I had to design the delivery flow around the real failure points of programs like this, which are not distance but coordination: pantries have specific intake hours, produce must move within about 48 hours of harvest, and someone has to actually claim the drive. Building verification by weight into the flow mattered, because it turns feel good numbers into numbers a masjid can show its donors.

Accomplishments that we're proud of

I am proud that the AI in my demo is real. Bustan makes live Claude API calls, knows what is planted in each bed, and gives genuinely useful advice, not scripted responses. I am proud of the delivery flow, because it answers the hard operational question of how food actually gets from a lawn to a pantry shelf, with a verified weight at the end. I am proud of the Barakah Tree as an emotional centerpiece. In testing, watching leaves appear after a delivery is verified is the moment people smile. And I am proud of grounding a modern product in a teaching from fourteen centuries ago in a way that feels natural, not forced.

What we learned

I learned that a good idea becomes a real product when it gets narrow and countable. My pitch went from a vision about gardening and charity to a single equation: one bed, 60 pounds, 50 meals, and 1,800 masjids that already run food programs as the warm market. I learned that the best products for faith communities do not bolt religion onto an app; they find where the tradition and the technology already want the same thing. Islam already had the concept of charity that keeps giving, and I just made it visible and verifiable. I also learned practical lessons about prompt engineering for a focused agent, since giving Bustan the garden's real context mattered more than clever wording.

What's next for Sadaqah Garden

Next, I want to pilot with two or three local masjids and their nearest pantries to run real harvest delivery logistics, with masjids paying a simple software subscription of about 75 dollars a month for volunteer scheduling, Bustan, pantry matching, and donor facing impact reports. I want to add plant diagnosis from photos through Claude's vision capabilities and give pantries their own accounts to post live produce needs. From there, the same playbook extends to churches, schools, and community centers with unused land, and to seasonal campaigns tied to the Islamic calendar, such as Ramadan iftar gardens and Eid harvest drives, because feeding your neighbor grows something good in every tradition.

Built With

• anthropic-claude-api
• css
• ios
• javascript
• react
• svg