News: paretos secures

4M Dollar Seed funding.

Learn more

Turn your data into
insightspredictionsscenariosdecisions

like a tech giant

20x faster results0$ integration costs100% business impactno data science skills needed

Customer recommendation

Recommendation engines are the heart of every modern infrastructure. Turn your static decision trees & fixed thresholds into self learning recommendations. Start to learn more around your customer.

Dynamic pricing

Static is so 1990s! Optimize the prices for your products tailored to customer cluster and react fast to changing markets by choosing the best trade-off between margin and buying probability.

Warehouse optimization

Reality is complex and fast changing. Do you know your future stock and how to steer best tradeoffs? Get dynamic predictions for your daily warehouse planning and do real time optimization.

Workflow

Connect

Easily connect your data with paretos using existing connectors. It enables automation and reduces the risk when exchanging data with other tools. For a smooth start, the platform of course also offers the upload of Excel / .csv files.

Insights

Get input/output relations and many other insights about your data. Paretos can show you, for example, whether you are looking at the right inputs or whether relevant parameters are still missing from your data. In addition, you can see at a glance the impact of your individual data on your business outcome. Become data-driven from day 1!

Predictions

Get predicted results based on specified inputs. With automated AI pipelines, we take care of selecting and training the best model for you and your data, and leave it to you to use for predictions. No need for “educated guess” anymore.

Trade-offs

Get all the best future scenarios and solutions at once and see all the trade-offs between the alternatives. Pareto’s AI based algorithm Socrates is designed to not only find the trade-offs much faster than other solutions but also to find many more optimal solutions.

Deep tech under the hood

Self learning

A self-learning, dynamic learning layer with over tens of thousands of hours of training enables paretos to deliver individual models and trade-offs for each problem.

Using numerous underlying comprehensive libraries of state-of-the-art AI models and optimizations, the paretos platform automatically selects and applies the most appropriate ones for the customer data.

FAQ

What kind of data do I need?

In general input – output data are required. Each input and output should be collected ideally in a separate row. Input data are defined as all data that may have an impact on at least one output data. E.g. marketing ad spends as input data for conversion rate. Input data can contain inputs on which you can have a direct impact e.g. ad spends or inputs which you can not influence but may have an impact on one of the outputs e.g. competitor price.

Is a minimum of data required?

Yes, but it might be less than you think. Some of our customers started with 100 rows of data. Our algorithms pick the right prediction model for your data automatically and trains it. Based on the training you get insights about the models and if it is enough data. Furthermore, our solution lets you know after adding iteratively new data if the models are improving or if you should consider to collect other inputs which could be relevant.

How long does it take to get started / first results?

After 2 weeks you receive the first insights about impacts of your input data on your outputs. The insights are generated based on a first prediction model. In case the data require big amount of pre-processing it may take longer depending on the data complexity.

Can I use existing models to get directly trade-offs?

Sure, when a model already exist the first results can be obtained within hours. Simply connect the model with paretos and get your dynamic trade-offs. In this case the model will run on your computing resources and only meta data are exchanged with paretos platform.

What others say

To connect our recommendation & prediction engine to the paretos API was super smooth and done in 2 days. Now our model is constantly improving with an AI based algorithm - this is a gamechanger for us and especially for our clients.

Ilhan Scheer

Managing Director DACH

Accenture

Paretos provides a perfect platform to speed up our activities around predictive analytics and dynamic pricing. We really enjoyed the smooth onboarding and the visualization of trade-offs to learn and iterate fast on our side without the need for more resources

Konrad Heimpel

VP Product & Data

Getsafe

Paretos stands for an approach to optimization that is focused on creating non-trivial insights. The optimization algorithm is designed for complex problems. It enables the discovery of better solutions with fewer evaluations

Markus Gürster

Project Lead Torpa

MIT

Leadership

Thorsten Heilig

Founder & CEO

Thorsten has a passion for how technology can solve complex (real-world) problems on a broader scale. He is always striving to find ways so that new methods can be used by “everyone”. With a focus on organizational and product growth, he has co-founded companies, worked as a management coach, and was most recently COO at moovel/REACH NOW before co-founding paretos. His motivation is to democratize AI-based technology to make better strategic decisions.

Fabian Rang

Founder & CTO

Fabian is fascinated by the possibilities of mathematical methods to handle complex data structures and models. Growing up as an engineer at a large German OEM, he chose systems engineering and mathematics - his PhD thesis is about “Efficient Multicriteria Optimization for Expensive High-Dimensional Blackbox Problems”. He is co-founder of paretos with the goal of bridging the gap between mathematics and business applications.

Éric Värnild

VP Engineering

Eric decided to join paretos when he saw how it succeeds in combining machine learning and optimization to deliver real business value. With 30 years of experience in software engineering and artificial intelligence in several management positions (e.g., at IBM or Rapidminder), he brings the expertise needed to hire a world-class technical team that can build world-class enterprise software as a service. He is committed to solving complex problems for clients using math and technology.