Meet European bootcamp Rockstart’s first AI cohort

Back in April the Netherlands-based Rockstart bootcamp announced it was adding an AI track. It’s now named the first batch of startups in the inaugural six-month program, with teams hailing from as far afield as India, Singapore, South Africa and the US.

Nine teams in all are on what Rockstart bills as Europe’s first AI accelerator — including a startup building an autonomous cleaner robot (pictured above), a smart parking spot finder and a platform to help analyze drone inspection imagery to help spot signs of damage to high value assets like wind turbines.

Selected teams each get €20,000 cash (in exchange for, typically, 6% equity) and €80,000 of “in-kind funding” — which Rockstart tells us can include incorporation, IP protection, mentoring, events, professional services (legal, accounting) etc, and also refers to the value of the office space in Den Bosch where startups work for the duration of the program.

The bootcamp kicked off in October, a little later than planned, as Rockstart ran a Launchtrack AI program in September for (very) early-stage startups — one of whom, GeoSpark, ended up getting into the full AI accelerator program.

Here’s a quick run down of the selected teams:

AirSquire (Singapore)
Bills itself as an “intelligent inspection management” startup, targeting construction managers and sales forces in real estate with tools such as a depth camera for capturing 3D indoor space and services aimed at automating marketing.

AizoBot (India)
The team says it’s drawing on experience building software for self-driving cars in India to devise its own autonomous robots for cleaning, collecting, and disposing of garbage and dust from streets and floors on their own. The bots are electric and cloud-connected, with the aim of making cleaning practices “green, consistent and easy to manage”.

Baarb (US)
A recommendation system for travelers to book hotels which says it analyzes hundreds of millions of data points, structured and unstructured, to serve up personalized results. It offers an API for booking platforms so they can feed in inventory and add personalized search to their sites and apps.

Birds.ai (The Netherlands)
A team that cut its teeth making a smart drone system for wildlife protection has here shifted focus to automated visual inspections for high-value assets such as wind turbines. Its cloud service analyzes images collected by drones to flag up and pinpoint damage or a defect.

Cambridge Humanae (Italy)
This startup is combining psychological tests and AI to try to understand human reliability by estimating the stress resilience of high-risk individuals, such as soldiers, pilots and policemen. It’s developing a SaaS monitoring platform with risk-tailored psychological tests plus AI — aiming to increase the accuracy of monitoring over time.

GeoSpark (India)
GeoSpark (previously known as Holo) is a startup born out of a problem encountered via the earlier location-based messaging app idea. The founder is now seeking to apply AI to tackle the problem of battery drain during location tracking, and reckons the tech can result in a 90 per cent reduction of an app’s energy consumption for location tracking by using predictive AI learning based on the past behavior of a user.

OPTOSS AI (The Netherlands)
OPTOSS AI is a tool designed for telecom and service providers operating critical networks and aims to augment human operator capabilities and help them prevent major incidents (e.g. network outages), as well as also offering protection against operational and cybersecurity threats.

Overscore AI (South Africa)
Overscore AI is playing in the natural language processing space with the aim of providing an automated screening engine for job applicants — using a proprietary algorithm and workflow stack they say is designed to “read” a document as a human would. The tech is looking at the relationships between words and then teaching the prediction engine to understand words in their context.

Peazy (India)
Peazy (aka Park Easy) is a parking search engine that uses a parking prediction model (plus a learning algorithm continuously gathering data from users) which applies machine learning on crowdsourced GPS data to (aim to) locate free parking lots.