logo

Get in touch

Join the age of practical AI

We build end-to-end AI solutions that leverage existing data to its fullest potential

Anthill

AI Consulting shape

and ML Development

a team of AI technology experts specializing in AI consulting and developing ML algorithms

As part of Anthill we naturally extend our software services with proprietary AI solutions

Anthill

Awesome Image Awesome Image Awesome Image Awesome Image Awesome Image

Case
Studies

ant-coe-case-study1-img1
The Client
AMPECO
  •  Software for management of EV charging stations
  • Funded by BMW iVentures
  •  Operates 4.5M+ charging sessions
ant-coe-case-study1-img2
The Challenge
What we had to solve
  •  Charger users anxious about station availability
  •  Station crowding at peak times
  •  EV Station operators cannot set dynamic rates and plan repairs
  •  Planning and balancing issues for grid operators
ant-coe-case-study1-img3
Our Solution
How we solved the problem?

Predict Usage of EV Charging Stations with AI

ant-coe-case-study1-img4-new
Our Approach
How we solved the problem

State-of-the-Art Algorithms

 Build upon existing academic AI research with papers as recent as 2022

Hybrid Model

 Combining domain knowledge and AI raised prediction accuracy from 70% to game-changing 85%

Minimally Invasive

 We accessed only the necessary data and integrated the solution without disrupting existing processes

ant-coe-case-study1-img5
Created Value
Our contribution
For Station Users:
  •  Seamless EV charging experiences
     Efficient trip planning

For Station Operators:

  •  Maximize profits through dynamic rates
  •  Plan repairs during downtime
ant-coe-case-study1-img6
Outcome
The end result
  •  Solution leveraged customer’s existing infrastructure and partnerships
  •  85% prediction (business applicable) accuracy for stations in Europe
  •  MLOps retraining pipeline to handle data drift* and ensure long-term performance
ant-coe-case-study2-img1
The Client
SCHÖELLY
  •  Top 10 medical device company
  •  High-end German optical devices
  •  Global reach
ant-coe-case-study2-img2
The Challenge
What we had to solve

Traditional QA:

  •  Manual inspection of endoscopes for blurring, contamination, misalignment

Challenges:

  •  Time-Consuming
  •  Subjective
  •  Costly
ant-coe-case-study2-img3
Our Solution
How we solved the problem?

AI-Powered QA for Production of Optical Devices

ant-coe-case-study2-img4-new
Our Approach
How we solved the problem

Hybrid Algorithm

 Combining traditional computer vision with domain-specific measurements

Integration of MTF

 Modular Transfer Function for measuring endoscope blur compatible with manual inspection

In Line with ISO13485

 ISO protocol for medical devices – 2000 page compliance validation

ant-coe-case-study2-img5
Created Value
Our contribution
  •  Faster QA
  •  Eliminated human bias
  •  Recall/Rework reduction
  •  Automated compliance
 
Used in the USA, Canada, and China
ant-coe-case-study2-img6-new
Outcome
The end result
  •  Benchmarked against industrial-grade software
  •  100% replicability through ISO-mandated testing
  •  In-Browser proprietary solution in line with medical regulations
  •  On-Premises Hosting avoiding international compliance problems

Technical
Stack