ARTIFICIAL INTELLIGENCE

We don't hype it.

We build it for our customers.

Learn more

Labra AI is a partner for companies who aspire to extract the maximum value from their data.

Services

AI BLUEPRINT

1 -2 weeks

The planning phase. We exchange domain knowledge with the client, identify AI opportunities and create a roadmap for the most promising opportunity.

EXCHANGE KNOWLEDGE
IDENTIFY OPPORTUNITIES
CREATE ROADMAP

AI SOLUTION

1 -2 months

The development phase. We gather the necessary data, build a proof-of-concept and iterate it into the AI Solution. The collected data can be anything - for example images, text, numbers or signal data.

GATHER DATA
DEVELOP PROOF-OF-CONCEPT
BUILD AI SOLUTION

AI INTEGRATION

Case-by-case

The use phase. We build the data pipeline, integrate the AI Solution and maintain it.

BUILD DATA PIPELINE
INTEGRATE AI SOLUTION
MAINTAIN AI SOLUTION
Book a meeting

Cases

Create a new service

Our client had envisioned a new predictive service for their platform which would speed up their customers' process and increase safety. For our proof of concept we designed an LSTM neural network and created a novel process activation detection algorithm. The network used sensory data to predict previously unknown process values. We delivered excellent results, however we cannot go into further details as the work is currently under a patent survey.

Accuracy +/- 3% error margin
%
Accuracy +/- 10% error margin
%

Segment images

One of our founders researched and published a paper on cell nuclei segmentation at Aalto University. The output of the research was a novel ensemble model combining the predictive power of multiple different neural networks. The ensemble model achieved state-of-the-art results leading the paper to be accepted into International Symposium on Biomedical Imaging.

Predict failures

Our client was experiencing problems machining hard metal which lead to unstable production and inefficiency. We were tasked with predicting breakdowns for the milling machine to allow the machine operator to carry out tasks instead of monitoring the machine. We used automation and sensory data to build software which predicts breakdowns in real-time and warns the operator in advance. Next we will roll-out similar software in other operations in the company.

Average number of failures predicted per month
0
% of failures predicted
%

References

Founders

Martti formulates business problems into artificial intelligence specifications and builds proof of concepts.

Martti Pankakoski

CEO


+358 50 495 0303
martti.pankakoski@labra.ai

Oskar, whose background is in machine learning research and software development, builds state-of-the-art artificial intelligence.

Oskar Vuola

AI Engineer


+358 40 502 2009
oskar.vuola@labra.ai

Johannes, whose experience ranges from electronics design to software development, integrates hardware and software seamlessly together.

Johannes Kaunisvaara

Automation and Robotics Engineer


+358 40 844 8080
johannes.kaunisvaara@labra.ai

Ville understands the clients' business needs and enables frictionless collaboration between the client and Labra.AI.

Ville Juutilainen

Business Developer


+358 409 654367
ville.juutilainen@labra.ai