1 month ago, we kicked off the Vapi Build Challenge with a specific question in mind: what happens when you give hundreds of developers a production-grade voice platform?
The goal was simple. We wanted to see what our community would ship if we gave them the tools and a reason to build.
The submissions confirmed our thinking. We saw agents deployed in active recruiting pipelines, tools for field technicians, and new interfaces for creating software itself.
After reviewing more than 300 submissions, here are the winners of Vapi Build Challenge 2025.
Real-World Applications
These projects solved specific, high-friction problems and were built for immediate use in real-world workflows.
- Lavar
- Problem: Reading is a passive activity. Taking notes or looking up concepts breaks your flow.
- Solution: An agent that reads books with you, remembers your questions, and logs your notes in context. It functions as a dedicated research assistant embedded directly into the reading experience.
- Talvin
- Problem: High-volume recruiting is manual, repetitive, and slow.
- Solution: An AI recruiter, already in production, that handles specific hiring workflows with precision. It automates initial screening, freeing up human teams to focus on qualified candidates.
Creative Use Cases
These agents impressed us with their originality, revealing entirely new ways in which humans might interact with computers.
- FROSTai
- Problem: Field technicians need access to complex information while their hands are occupied.
- Solution: A ruggedized support agent that provides hands-free instructions and guidance. It’s designed to fit the exact workflow of a technician on a job site.
- Jiffly
- Problem: Finding and booking local services requires multiple phone calls and manual coordination.
- Solution: A consumer concierge that handles outbound calls to find and book appointments for you. It’s a practical step toward a future where agents transact on our behalf.
Community Picks
These projects gained significant traction for their clear execution and thoughtful application of voice AI to solve a single, well-defined problem.
- Adulting Buddy: A clean, mobile-native personal assistant for managing daily tasks and staying organized.
- JobGenie: A focused recruiter agent designed to help students land co-op jobs, with a conversational flow aligned to its mission.
- Candling: An agent for first responders that triages emergency calls, escalates crises, and auto-generates case summaries to reduce administrative load.
- Vapinception: An agent that lets you create other agents using only your voice, turning a complex configuration process into a simple conversation.
- LeadSense: A sales agent built to handle outbound calls for car dealerships, targeting a high-value, fast-growing category for voice automation.
The Path Forward
This challenge gave us a clear signal of where voice AI is heading. The focus is shifting from simple conversation to concrete execution.
Whether it's reliably handling thousands of daily support calls or guiding a technician through a complex repair, the goal is the same: building agents that complete a job.
The winning projects show this principle in action.
They take the core capabilities of voice.. understanding, responding, and acting.. and apply them to specific, high-leverage problems.
This is the next phase of voice AI- less about a single, all-knowing assistant and more about a distributed network of agents executing real-world tasks.
Our job is to provide the reliable infrastructure for that network. And your job is to build the nodes.
The work continues.
