Education

Two lessons learnt to take your AI from idea to production

Are you interested to start using artificial intelligence (AI) in your credit underwriting business? Here are two lessons that I have learnt, which might help you convince your organization to use AI.

Artificial Intelligence improves credit scoring, the question is how much the impact will be on your business?

It is no secret that AI improves credit underwriting. How much it will affect your business is impossible to answer before trying it –  since it depends highly on the AI technology and what data is used.

There exist several options to understand how much AI will improve your specific business. One classic method is to run a proof-of-concept. The proof-of-concept path involves building an AI model, comparing the AI model performance to your current model, and analyzing the incremental business impact that the AI model delivers. If you are experienced within AI you can try it out yourself, you can hire people that are experts within this field or sign-up for our pilot study. Either way, you will ultimately discover that AI underwriting will add significant value to your business.

Deploying the AI model into production is the hard challenge

There are several challenges to accomplish before deploying an AI model into production. Including IT-Integrations, regulatory and compliance issues. In my experience, we have found suitable, low-friction IT solutions that are agnostic to whether a lender prefers to run its solution on-premises or in the cloud.

The real challenge is model transparency, a credit underwriting model has to be understood so it can pass your existing risk management committee and compliance requirement. This requires you to dig deep into the complex math behind the AI model –  my recommendation is to find a partner who already has addressed these types of challenges before since it is really time-consuming, math-heavy and complex process. At Evispot we have developed “Evispot Traits”, a solution customized to give creditors the transparency and explainability required to take an AI model into production.

If you’re ready to capture the benefits of more accurate and transparent credit risk models, let’s have a chat! Contact partnership@evispot.ai to learn more or to sign up for a pilot study.

read more
Tomas SelldenTwo lessons learnt to take your AI from idea to production

… and the winners are! Evispot AI Challenge

Most Accurate Solution 


1. EZW (70,3%)
By focusing on the core problem EZW was able to developed a superior solution compared to the competition. EZW had a structured pre-processing and used random forest algorithm which resulted in the winning solution.
You now have the right to brag about your machine learning skills! 

A big congrats, well deserved! You are now going to tell Bisnode how they should implement machine learning.

2. Real Human Beings (69,1%)

Well done, Real Human Beings! A well deserved second place! Congratz. You will get azure passes!

3. Klubb Bubbel (64,7%)
Congratz! You will also get azure passes!

 

Most Innovative Solution 

Deep Engima
Many of the solutions followed a similar processes in terms of pre-processing and choice of algorithm – but Deep Enigma stood out in terms of application area. Deep Engima is winner of the most innovative solution – by proposing a suitable & creative application area of your solution! Congratz!

You are winning a night of beers with Evispot – let’s talk about how to develop your proposed solution!


Once again, a big thanks to Mikael Kågebäck, Christian Lauritzen and David Fendrich in the jury and to our partners! 

Congratulations to the winners! 

read more
Tomas Sellden… and the winners are! Evispot AI Challenge

Evispot AI Challenge: Powered by Bisnode – has now started!

Yesterday, Evispot kicked off the Evispot AI Challenge, a five day long machine learning competition, co-organized  with Bisnode and Microsoft. With just over 20 teams participating in the challenge Evispot invited the teams to a night of inspiring talks, food & drinks and introduction to the challenge itself. Besides Evispot, we listened to talks from David Fendrich of Crawlica under the theme “Why can we predict anything at all?” and Robert Quinn of IBM under the theme “Did you say Let’s eat Grandpa or Let’s eat, Granpa?”. Now the participants have until Tuesday to complete the case – we are looking forward for the results!

 


Photo: Fabian Wennerbeck 
Joachim Karlsson, Group Director of Innovation, Bisnode

Joachim introduced Bisnode and their view of the Evispot AI Challenge as well as highlighted the importance of machine learning in the future of credit decisions.

 

 


Photo: Fabian Wennerbeck 
”Why can we predict anything at all?” by David Fendrich, CTO Crawlica

David started off by saying that he would give a technical speech, not to talk about business at all. And he certainly did, David speech focused around the Solomonoff induction, Kolmogorov Complexity
and gave some valuable tips of how to tackle AI problems. You can find David’s slide here. 


Photo: Fabian Wennerbeck
“Did you say Let’s eat Grandpa or Let’s eat, Grandpa?” – by Robert Quinn, Senior Software Engineer, IBM

Robert talked about how IBM Watson cognitive journey started with playing Jeopardy. By building a Jeopardy AI-robot, with the two requirements, it should answer within one second and at least be 85% sure to answer the correct answer.


Photo: Fabian Wennerbeck 
Isak, CFO & Co-Founder, Evispot
Isak highlighted the importance of the challenge due to the upcoming regulatory changes in the financial industry.




Photo: Fabian Wennerbeck 
Tomas, CTO & Co-founder,  Evispot
Tomas introduced the challenge in details and declared the challenge’s start. 

 

Below are more pictures from the evening.


Photo: Fabian Wennerbeck 

Photo: Fabian Wennerbeck 

Photo: Fabian Wennerbeck 

A big thanks to everyone who showed up and had a great evening with us! If you have any questions or comments don’t hesitate to contact us – hello@evispot.ai 

Now we are looking forward for the results!

A special thanks to Fabian Wennerbeck who took the pictures!

read more
Tomas SelldenEvispot AI Challenge: Powered by Bisnode – has now started!