Introducing 
Express — Go from prompt to pipeline in minutes.
This past quarter, Nexla took Express.dev on a hackathon road trip – from Bangalore to San Francisco. Across them, teams built production apps based on real problems, using real data, and put Express.dev to the test. The results showed what happens when builders have an idea and use Express.dev to power the apps. Here is a look at the standout projects from each event.
Pet medical data today is scattered across paper records, PDFs, lab systems, prescriptions, and visit notes. There is no single view, no easy way to spot risks like drug interactions, and nothing clean you can hand to a vet.
The winning team built a web app that centralizes all of that into one timeline, runs it through Express.dev for AI-powered analysis, and flags potential risks. The output is a downloadable, printable report that can actually be used in a real clinical setting. It was a clear example of working with messy, real-world inputs and using Nexla to streamline it the way it is meant to be used.
See the PetMed Timeline App in Action
Founders and small teams often lack real-time visibility across their business. Metrics live in one tool, system health in another, customer tickets somewhere else and problems are usually caught too late.
This team built a centralized platform that monitors KPIs, microservice health, and customer issues in real time, letting founders ask questions in natural language and get immediate, actionable answers. What made it stand out: Express.dev was embedded throughout the entire stack – from connectors, transforms, schemas, and workflows all ran on it.
Check Out the Centralized Metrics Platform on LinkedIn
Vyora, an AI-powered sponsorship intelligence platform for YouTube creators. The app analyzes any public YouTube channel and surfaces data-driven insights: what brands are paying creators like you, which brands are actively sponsoring, and how you compare to similar creators in the market. Express.dev powered the data ingestion, transformation, and transfer layer underneath it all.
Experience Vyora + the Power of Express
A market sentiment engine that analyzes how individual news events might affect the price of a particular stock. The project pulled data from GDELT, a free, open, global database of news events, and used Express.dev for seamless integration with Google BigQuery, enabling fast data ingestion and analysis at scale.
See the Market Sentiment Engine on X
A scholarship discovery platform for students using web scraping powered by Express.dev. The app was completed in under two hours and demonstrated how quickly Express.dev can be used to gather, route, and surface unstructured web data to power a live production app.
See How Web Scraping with Express Works
Monitoring the Situation is a real-time situational awareness dashboard for San Francisco. The team described the inspiration this way: “San Francisco is loud – but only if you know how to listen.”
Built in a single hackathon session on a MacBook Air, the app decodes P25 trunked radio from SFFD, SFPD, EMS, and mutual aid channels, transcribes audio in real time using faster-whisper, and fuses it with flight tracking data from FlightRadar24, live bus positions, and city dispatch records from DataSF – all rendered onto one living map.
Gemini geocodes raw radio transcripts (such as “units responding to 16th and Mission”) into exact map coordinates in real time. The entire pipeline runs end-to-end in under 10 seconds from raw RF signal to a geocoded, AI-enriched incident on the map. A standout moment: Gemini autonomously correlated a radio transcript, a FlightRadar24 police helicopter, and a DataSF dispatch call into a single unified incident without any explicit instruction to do so.
See the Real-Time Awareness Dashboard
Built in a single hackathon session, ForgeRedemption is a multi-agent game where each agent has its own role, skills, and decision logic. Five sequential LLM calls fire per turn, each reading and writing shared world state. The team built a custom dispatch layer to prevent agent conflicts, a JSON extraction fallback so the game never stalls on a malformed response.
Four AI agents break out of prison – orchestrated by Claude on InsForge. The team that built this searched the web using TinyFish. Called Express.dev from an edge function, which pushed the web scraped messy data into a webhook source and then did Python transforms and sent updated skill.md from nexset to pgvector, which were added as skill embeddings in a skill library in-game use and built RAG knowledge on the fly, and used VAPI to narrate it all live.
Check Out Forge Redemption’s Project Brief
Conversational data engineering platform for data pipelines. Describe what you need in plain English, Express builds it. Example: “Connect Salesforce to Snowflake, sync accounts daily” → pipeline generated in 3 minutes vs 3 weeks traditional. Try it free at Express.dev
Teams built applications like medical data timelines, real-time monitoring dashboards, market sentiment engines, and multi-agent AI systems.
Many teams built fully functional, production-ready applications within a few hours during the hackathons.
It simplifies working with messy, distributed data by providing connectors, transformations, and workflows needed to power AI applications. It gets you the data your agents need, so you need to only worry about
Enterprise AI agents fail when the context behind their decisions is incomplete, stale, or conflicting. Context engineering ensures agents receive accurate, permission-aware runtime context for reliable decisions.
Reusable data products unify databases, PDFs, and logs with metadata, validation, and lineage to enable join-aware RAG retrieval for reliable GenAI applications.
