Jaywing: Can the Company Outperform Tracer?

December 14, 2017

I am a person who gathers odd items of information. Not long ago, I learned about Thomson Reuters’ Tracer smart system for risk prediction assessment and other helpful services for those hungry for a data edge.

The idea is that wonks at Thomson Reuters figured out how to make sense of content in order to “predict” whether an investment poses a risk or a big trend will be, well, big.

From my point of view, a large outfit like Thomson Reuters suffers from size, management incentive programs which focus on specific financial goals, and technology lag. The term “technology lag” means that by the time a giant outfit decides a technology is important and useful, the technology is likely to be an old hat, a fedora from the 1950s perhaps?

I read about Jaywing, which is not a particularly easy thing to do. If you navigate to the Jaywing Web site, there are pictures and jargon which can be tough to decipher. For instance, the company is involved with “data science.” The company wants to delivery “strategy.” The company is “creative.” These are useful words, but I am not sure I can define them. I know that when Jaywing says, “Data science” and then explains it with by saying “It’s in our blood”, I think I understand, but I don’t really.

I did note an article in Global Banking and Finance which helps to explain what the company purports to do. “Jaywing Launches new AI Risk Technology Product Archetype.” I highlighted several statements about the interesting “AI risk technology” service; to wit:

Archetype uses deep neural networks, an advanced predictive modeling methodology and an intuitive user interface to create models that will radically improve the interpretability and predictability of the AI-driven scoring process.

Yep, easy model tweaking. When applied to fraud detection, Archetype works wonders.

I also noted this statement from the managing director of the firm:

Initial comparative tests of Archetype against an already strong fraud model have shown an uplift in the order of 10 points.

If I understand the statement, if a competitive product from one of the FinTech vendors delivers 85 percent accuracy, Jaywing pumps out an an interesting 95 percent accuracy. Now, if true, that is significantly better than outfits like Google can deliver. Most smart software chugs along with 80 to 85 percent accuracy. Quite a significant leap that 10 points makes. I can’t verify this performance, but it suggests that Jaywing knows how to catch one’s attention and financial fraudsters.

I circled this statement from Jaywing’s CEO (Rob Shaw) too:

We have now brought new algorithmic and expert knowledge on machine learning and AI to further optimize risk management, regulations and operational control while freeing resource to focus on other value-adding tasks.

Those blue chip, McKinsey like consultants may find themselves losing engagements to Jaywing if these assertions are on the money.

The big question is, “Is Jaywing able to deliver a significant boost in accuracy?”

Worth watching. I assume Thomson Reuters, among others in the smart software for risk analysis game, will be paying attention to the UK company. If the existing vendors can’t match Jaywing, I am 85 percent confident their marketing departments can deliver when explaining the benefits of their systems.

Jaywing’s intelligence unit states,

“We free marketers to think, not do.”

Competitors may want to note that Jaywing operates Epiphany and Bloom, two marketing companies. One of these firms asserts, “The marketing agency where artificial intelligence meets human creativity.”

When pitching high accuracy smart software outputs, marketing is important. Who wouldn’t like predictions that are 95 percent accurate?

Stephen E Arnold, December 14, 2017


One Response to “Jaywing: Can the Company Outperform Tracer?”

  1. Martin Smith on December 15th, 2017 11:06 am

    Hi Stephen, I have just come across your article online and wanted to drop you a quick note. Thank you for taking the time to look through our site. I can see what you mean – we have a very complicated business encompassing both data specialists and creative experts. I look after product development for Jaywing so can talk you through it.

    Overall, I’d describe us as a creative marketing company with strong brand and digital advertising capabilities, as well as data science credentials (and a data consultancy function) which sits at the heart of the company – hence the mix of descriptions on our website. In our data science arm, we do modelling and analytics for financial services and marketing teams – anyone who wishes to make better use of the data they hold. Archetype is primarily designed for credit, risk, fraud and marketing propensity modelling – instances where you’re trying to predict a specific outcome (such as a credit loss) from sets of historical data. Its key advantage is that it uses artificial intelligence to automate modelling processes that would currently take several days, creating accurate models in minutes. In developing Archetype, we’ve invented a methodology that overcomes a problem that has prevented the widespread take-up of AI in credit risk modelling: the requirement for such models to be explainable, transparent and to follow common sense rules. Archetype guarantees to produce models that do this. We think this is a game-changer.

    You can learn more about our product here: https://risk.jaywing.com/specialisms/artificial-intelligence/archetype/



  • Archives

  • Recent Posts

  • Meta