Mabe Flow / AI & apps
An HR assistant that keeps the conversation on device.
Mabe Flow is a native HR platform built for the Enactus–Apple competition, combining private assistance, semantic search, and predictive ML.
- My role
- iOS / AI development · hackathon team
- When
- May 2026
- Built with
- Swift · iOS · Apple Foundation Models · Tool calling · Semantic search · Machine learning
Explore how it works
Try it yourself ↓A request starts with a draft you can review.
Choose vacation dates, compare them with a sample balance, and review the prepared request.
18 allocated · 6 already used
The prototype counts both dates and every calendar day between them.
7 sample days would remain.
A local illustration of the vacation-balance and request-drafting tools. Days are counted inclusively, as in the prototype. No on-device model runs, and nothing is sent to HR.
The challenge
Employee self-service can require navigating scattered enterprise information. Our challenge was to bring useful assistance into a native mobile experience while keeping the AI private and grounding its answers in cited data.
How I built it
Keep assistance local
I built with Apple Foundation Models to support on-device assistance in a native iOS HR platform for Mabe.
Connect answers to information
Tool calling and cited enterprise data gave the assistant a way to work with relevant information and show where its answers came from.
Go beyond a chat box
I developed semantic search and predictive ML to help employees find information and estimate workflow outcomes.
What came out of it
Our team won first place nationally at the Enactus Change Makers Hackathon 2026 in Mexico City, an Enactus–Apple iOS development competition.
Created as a competition project for Mabe.
Sources & project context
More from the project
