Overview Your prompt What we build MongoDB Atlas + AI MongoDB Atlas Agent Engine Memory SDK At the kiosk How we built it
BUILD LAB ASSISTANT Ask a question before you build ↓
One prompt.One live app.
One prompt becomes a deployed web app with a MongoDB Atlas database backend and an optional built-in LLM chosen for the job.
BUILD · EXPLORE · VOTE Enter your prompt at a kiosk or scan the build QR code to use your own device. Then explore what has been built and vote for your favorite. Find a MongoDB staff member whenever you have a question.
01 PROMPT Tell us what you want to build
02 GUARDRAIL REVIEW Safety checks run before planning
03 REVIEW PLAN Choose feedback or submit
04 WE BUILD IT Your submitted plan enters the build queue
AVAILABLE INGREDIENTS Your app can use much more than text and forms. AI features — summaries, recommendations, comparisons, writingSearch — keyword & meaning-basedImages — upload, camera, AI vision, image searchDocuments — upload & analysisMaps & location — weather, directionsCharts & trends — data visualizationMongoDB Atlas Agent Engine Memory SDK — durable, user-specific context across conversationsWeb search — sourced starter data and live discovery when current public information matters
HOW YOUR APP IS SCORED 100 POINTS
40 MONGODB + TECHNICAL AMBITION Use MongoDB Atlas thoughtfully and propose capabilities that fit the idea—not product name-dropping.
25 IMPACT + UTILITY Solve a real problem for people, communities, organizations, or an industry.
20 CREATIVITY + ORIGINALITY 15 points measure originality against distinct prompts from past Build Lab events using cosine similarity of the full prompt, a concise concept, and the defining interaction. The score balances the closest match, the five closest matches, and the number above a calibrated similarity cutoff. Familiar ideas repeated across many projects score lower. Another 5 points judge the creativity of the interaction. Historical comparison prompts stay private.
5 OBSERVED BUILD RECOVERY Score the app from its actual validation and repair history: first-pass success earns 0; progressively deeper successful recovery earns up to 5.
10 LIVE IMPLEMENTATION After the build, a working, clear, reliable app earns the final implementation points.
Your idea earns up to 85 points. A verified live build adds up to 5 points for observed recovery challenge and 10 points for implementation quality.
Build from anywhere Scan the build QR code to open the unnumbered, untimed prompt page on your own device.
1: Scan the QR code.2: Prompt your app.3: Find MongoDB staff for questions.
READ ANYTIME · PUBLIC BUILD LAB GUIDE Everything shown at the kiosk is online. Scan for the intake fields, review path, sharing choices, data limits, scoring, safety boundaries, and prompt help.
OPEN KIOSK GUIDE →
INCLUDED
What the builder delivers.
Product plan. We first confirm the architecture can deliver the core idea, then run the shared safety review. After both checks, you can review and revise the focused build brief before building.
MongoDB Atlas data model + LLM. We design the document model around how the app reads and writes data, choose the MongoDB Atlas features, and select the right LLM. When the app is ready, its Apps card shows the live collections, field types, and embedded structure.
Build + test. We create the interface, connect every control, test the real data flow, and repair problems.
Production deploy. We publish the finished app. After you submit, you get one email with a private tracking link and a second email when the app is ready. The ready page includes one optional post-build feedback round.
Your prompt is the only technical setup you need. No database configuration, model selection, API keys, coding, or deployment steps.
ASK ABOUT YOUR BUILD Questions before you build. Ask about writing the prompt, supported capabilities, MongoDB Atlas and AI, data limits, safety, sharing, or what happens after submission. This assistant answers questions based on the guides above; it does not submit or create an app.
YOUR FEEDBACK
Help us improve the guide.