---
post_type: "project"
title: "IIA (Intelligence Inventory Assistant)"
blurb: "A camera-first mobile app for managing collections — point it at an object and the app recognizes it and fills in the details."
org: suslib
year:
    - 2020
tags:
    - AR
    - ML
    - inventory management
    - mobile app
stack:
    - ARKit/ARCore
    - Knowledge Recognition API
    - TensorFlow
    - Elasticsearch
    - REST API
link: "https://suslib.com/solution/inventory-assistant"
---

IIA started as a side experiment to see what our Knowledge Recognition API could do in practice, and turned into a full product. You point a tablet camera at an object in a collection and the app tells you whether it's already in the inventory, lets you check it in or out, or adds it with the metadata filled in for you. The camera was the whole interface — no forms, no RFID scanners. Underneath it used our KR API: Word2Vec and Elasticsearch for semantic search, computer vision to identify the object.

I led the product engineering through several iterations, from a concept trained on our team's own bookcases to web prototypes and then a beta people could register for. We shipped version 1.0 on iOS and Android in 2022. It was the first time we turned our research into something real people used to manage collections, and the part that stayed with me was the interface decision: let someone act by pointing a camera instead of asking them to learn a form. Changing how we interact with software, so it adapts to the person, has been my work ever since.

`with` [Martijn de Heer](https://suslib.com)