The ability to react to small business capital needs, and to underwrite it in an efficient way, is challenging, and that's what I think we have learned to do well with cash flow underwriting.

That's not to say that we don't look at the credit profile of our customer, we do, but it's a much smaller percent of our proprietary scoring models than the cash flow.

Do you know one person in Canada? Then, if you put all the names of the entire population of Canada into a hat and draw one at random, you're eight times more likely to draw that friend's name than you are to win the Mega Millions jackpot.

So it's so unlikely. But here's a simple question of probability: why are there winners?

Because there are winners of this jackpot, fairly often, not every drawing, but fairly often. And if the odds are so low, as I say they are, why are there winners?

One of the UK banks that we work with, they have 17 lending systems. That’s really complicated to work with. People can change. Technology can change. What doesn't change is the process. So you have to adapt to every banks process. Technology has to adapt to every process. So essentially, people adapt, and technology adapts. But the process is something which doesn't change, and that is something which is the learning process for everyone when you work with a particular institution.

AI is emerging as a as a transformative force, which is redefining banking: from digital lending to advisory, to research, it’s across all its different business units.

The innovation of our solution lies in a number of areas, but I'll speak about two key areas: and that is our awareness of the importance that AI plays today in risk decision making; and the importance that data plays in making informed decisions.

On the AI front, Provenir's auto ML product allows our customers to quickly easily and affordably develop new or improved credit risk scores, which are instantly deployable into those credit decisioning workflows on the Provenir platform.

.I think it was partly due to a chance that we eventually built Taran DM because it was at the beginning of 2020 and we had two interesting projects. So it was like a lot of potential and one of them was a decision management platform, like some customised one with scorecards for a new fintech. And the other one was a platform for a large German automotive company, which was supposed to optimise their supply chain. Middle of February we were flying to the German company, to the exporter. And we had a really good session with the management team, getting a lot of ideas on how to move it further. We were super excited about it, but as I said, like it was February 2020.

So when COVID was spreading within a few weeks, those guys stopped answering our emails and then we got back to them later, they said sorry, but our supply chains have gone haywire because of Covid, we cannot do anything a few months or maybe even a few years!

So that project got killed. And then we ended up with the other second big project, which honestly was, I think, a better fit for us because most of the team was coming from the finance, we had like the experience with infrastructures doing like real time decisions, whether it was credit risk, whether it was the high frequency trading was a large part of the team came from actual credit risk teams that may have been using the tools such as FICO Blaze or Experian Power Curve. We know the the strong points, the weak points, so big up to the I would say drawing board and we thought maybe there is like a market opportunity here or let's you know, let's basically build something.

CreditPy is including some functionalities regarding to develop credit risk scorecards, a PD model, basic data analysis, and it checks the informative variables in an automated way to determine which features is going to be passed to the predictive model. It also generates an automated model framework that is actually searching for the best predictive model across the different feature sets that potentially can be used during the model development.

And after this, there are actually many functions that has been defined to create the rating scale. And also, after creating the rating scale, its offers to do some validation, like univariate gini check, information value checks, basic multicollinearity checks, stability checks on the futures to see if there will be any drift on the predictions on the auto sample set bit applies a basic rate of evidence transformation on the data.

And finally, it allows the user to validate the created rating scale, predictive power of the model and the calibration.