Insurance
Economic forecasting for client advisors
An international insurer wanted its client advisors to have current economic forecasts at hand. We built models for key indicators and fed the results into the advisory tool they already use.
Data Science & AI Exploration
Forecasting, machine learning, GenAI and geodata models.
We build forecasting models, classifiers, GenAI applications and models that use geographic data. Most projects start with an analysis of your data and a proof of concept, then move to production.
The problem
Many companies collect plenty of data but have nobody with the time or training to model it.
Questions such as next quarter's demand or which addresses carry more risk stay unanswered.
How we work
- Initial Assessment
Review the available data and pick the questions worth answering first.
- Data Collection & Preparation
Collect missing data, clean it and prepare it for modelling.
- Exploratory Analysis
Statistical analysis to see which patterns are real and useful.
- PoC Development
Build a first model and test whether it is accurate enough to use.
- Solution Design & Handover
Design the production version and plan how to get there.
What you get
- Data analysis & visualizations
Statistical analysis with charts your business teams can read.
- ML/AI models in production
Trained, validated and deployed models.
- Business interpretation
A plain explanation of what the model outputs mean and where their limits are.
- Data measurement guidance
Advice on what additional data to collect and how.
- Knowledge transfer
Workshops so your team can run and maintain the models.
Projects
Consumer Goods
Demand and sales forecasting
A consumer goods company planned production mostly from last year's figures. We built forecasting models that take seasonality, promotions and external data into account. Planning now takes less time.
InsurTech
Customer care chatbot
An InsurTech wanted to automate routine customer questions. We built a chatbot on their own data and put it into production. Handling time for routine queries went down.
Manufacturing
Logistics efficiency analysis
A construction materials company suspected its logistics were costing too much. We analysed the operational data, located the bottlenecks and proposed specific changes.
Insurance
Spatial risk scoring with GIS
An insurer wanted to rate property risk by exact location instead of postcode. We added GIS data to their rating models on Google Cloud, which improved how well they separate high-risk from low-risk addresses.
IoT / Fleet
GPS telemetry reconstruction
An IoT fleet operator had millions of raw GPS points. We wrote algorithms that reconstruct the areas each vehicle actually operates in, so the operator can see how the fleet is used.
Write to Fredrik directly.
Fredrik Moeschlin CEO & Founder