Five AI startups impacting society for the better

Written by
Christiaan Wiers
Table of Contents

With the introduction of ChatGPT it looks like the era of artificial intelligence has begun, however the true story is that companies have been developing AI applications for many years already.  Next to the more familiar AI applications such as Siri (for iPhone users), Amazon’s Alexa or Deepl for translating texts. Many startups are trying to catch the wave of AI, even in areas where you wouldn’t expect AI to be found. In this article we present a selection of five unique startups which are using the power of AI to impact society for the better. This selection has been made in order to show how diverse and useful AI can be in helping the society move forward. But before showing these startups, some more details about the AI trend will be shown. 

Why are there so many new artificial intelligence startups?

The amount of startups focused on AI is booming, this has all to do with the increasing interest in artificial intelligence. As an indication for this increasing interest in AI, a search about on ‘artificial intelligence’ in Google Trends gives us a hockey style curved graph. Here the interest in this keyword is given in terms of searches as a percentage of the all-time high. 

Chart, line chartDescription automatically generated
Source: Google Trends

On top of the increasing interest in AI, a survey performed by Deloitte in October 2022 shows how 94% of business leaders surveyed agree that AI is critical for business in order to succeed over the next five years.

ChartDescription automatically generated
Source: Deloitte

With the importance of AI for future business success underlined, it is time to dive into rising AI startups you may have never heard of before. 

Venture IQ’s selection of impact-driven AI startups

This unique selection of startups has been made by only choosing from startups which meet the following two requirements, namely they should:

  • Help solve fundamental problems for society 
  • Differ from ChatGPT-like Neural Language Processing applications

Mapdwell: assessing any building for solar panel potential 

Mapdwell has developed a state-of-the-art rooftop solar yield forecasting application. Using their tool you can quickly discover the opportunities for solar PV on various buildings. The tool does so by giving users the options to select buildings upon which they would like a report about the potential of solar PV.  This report gets generated by the software and contains a solar yield, revenue and cost estimation. More specifically, the  report includes technical details such as the amount of m2 of solar panels and it’s electricity output in (kWh/yr) but also financial details like the estimated revenue, costs and payback time. 

The magic happens in these advanced estimations made by Mapdwell’s software since it takes into account very detailed factors like rooftop geometry, local weather and potential obstructions (e.g. trees and nearby buildings) without requiring any physical measurement by the user.

Mapdwell | Tag | ArchDaily

Fion Technologies: an AI powered simulator for predicting wildfire behavior up to 10x more accurately than existing models

With temperatures globally rising we see how both the number of wildfires is increasing as well as the area of land which gets burned (see charts below).

Chart, bar chart, line chart, histogramDescription automatically generated

Fion technologies has developed an AI-based simulator which can simulate the development of wildfires. Doing so allows for more effective firefighting and safety precautions. Next to the simulator the company also provides an app with a map function in order to improve situational awareness for firefighters. On the map, firefighters can share their locations and indicate dangerous areas.


MapDescription automatically generated

SkinVision: detecting skin cancer using an app

This app uses an AI model to detect skin cancer and it does so very effectively with over 1.8 million users and over 3.5 million skin checks, over 55 thousand skin cancers have been detected already. Their model has been trained on over three million images and currently has an accuracy of 95%.

The company explicitly states that the app isn’t a replacement for a general practitioner but they do see how an increasing number of general practitioners will actually advice people to first make a check with the app. Hence enabling the general practitioner to work more effectively and also underlining their confidence in this app. When looking at the statistics, around one in five people get skin cancer in Europe and two out of three in New Zealand and Australia therefore this app has the potential to save a lot lives in the coming years thanks to the power of AI.

1,2 miljoen mensen checken huid op kanker met Nederlandse app | RTL Nieuws

DuckDuckGoose: using AI to detect deepfake videos and images

With AI becoming more accessible, it does so for people with bad intentions as well. One example of such a bad intention is using deepfakes to trick people into believing they are some person (e.g. with authority) which in reality they are not. Doing so can have all sorts of bad consequences. 

DuckDuckGoose’s software leverages AI to tackle an AI application with malicious intentions namely deepfakes. Their software is able to recognize deepfakes with an accuracy of over 92% and the model actually provides insight into areas which the model believes are AI generated by means of a heatmap. The software comes with API-integration as well and therefore their deepfakes detection software can be easily implemented into a wide range of applications. 

Detecting Deepfakes
Source: helping biologists design improved proteins in record time using powerful prediction algorithms and AI design suggestions 

Cradle helps biologists design improved proteins in record time using their prediction algorithm for predicting 3D structures and their AI models in order to generated new sequences give a specific biological application. This way the company is helping biologists to overcome challenges such as high failure rates and long development times. 

More specifically the AI models can be used to improve the protein’s activity (producing more of something) as well as it’s themostability (the ability of the protein to maintain it’s properties at high temperatures).

Cradle’s angel investors include, among others, John Zimmer (CEO & Founder at Lyft), Feike Sijbesma (Honorary Chairman and former CEO at Royal DSM) and Adriaan Mol (Founder at Mollie & Messagebird).


All related tags to this article:
Article written by
Christiaan Wiers