What technology development seems to have gone beyond the hype cycle for FinTech start-ups?  We feel it is Artificial Intelligence (AI).  Dave has been using it for Cogo Labs partner’s companies to create efficient ways to serve clients’ needs in a variety of markets.  This cost effectiveness is due to the lack of human capital needed to assist the client.

We are also seeing it applied to big data problem sets similar to what we do to search the Cogo Labs data for information on possible investments for the portfolio.  Many of you know this is called the Apollo tool.  The bigger the data set, the more interesting are results from applying AI but the more difficult the process of getting results.

Recent applications include Micronotes who we invested in late in 2017.  They use AI to help banks help their clients with needs based solutions such as lending or wealth management.  We saw another plan recently for a lending platform driven by AI.

We wanted to dig a bit deeper on the topic to share our insights on the state of play for this tool.

—Dave & Mark

The Basics    

It’s a labor replacement.  This is the key factor to remember when presented with an idea or opportunity for AI.  It uses algorithms to sort through data and learn patterns and “answers” to questions one is seeking in the data.

Let’s use online employment.  You can review Linked In searching for types of employees to hire for your new analyst role.  You might search for those people with an analyst title.  You might search for firms you admire for their training of analysts so you can pluck them already trained by someone else.  Am sure you have done this type of search before and have realized how much time it takes to sort out the information.  You are the AI algorithm.
Instead you can use a search firm like Jobcase to perform these tasks within their database of candidates (tens of millions)and provide specs to drive candidate discovery.  That is AI being used to more efficiently do a task.

A further benefit to AI, especially in this example, is that it eliminates bias.   We all have a bias-it can be good or bad on the human level.  Maybe people who wear glasses look smart to you so you unconsciously favor those candidates with glasses.  You don’t experience the bias or you would fix it.  Try a test the next time you are searching for something see if you see a pattern in your choices and your non choices.  Perhaps you will see a bias and become conscious about it.

AI, when trained properly, will not have that same bias.  It can create it’s own mistakes; however, so it is not infallible.  Perhaps it is looking at data and favors candidates close to the office.  Seems an easy way to ask a program to sort candidates but it introduces a bias.  Maybe the neighborhoods within five miles to the office are diverse in candidate profiles or perhaps they are not.  Maybe that area is next to a university so many candidates come from the same school.  You would likely want to pick from many types of schools rather than one to find the very best choice but your data set and your algo are training on only this biased dataset.  Beware this automated form of bias.

AI is a wonderful tool but humans still need to think about the criteria and insight they need to automate.  They need to really dig into choices around data and AI tools to make sure they are getting the best benefit of the technology.
-Mark

Patterns Are Telling

I bet that high on your list of issues to think about for your organization is how to oversee its digital transformation.  I hear that all the time from large companies in finanical services.

Portfolio companies are an answer for them.  Partner with them to drive this change to digital. Find ways to experiment in the organization.  Do not be afraid to fail.

Our fun way of expressing how data scientists want to find patterns in data we have at our fingertips is “Disneyland for Quaints”.  As a result of having built such a resource, we spend time thinking about the governance of the use of the data for our purposes or in support of a portfolio company.  This structure is the same for you in your firm too.

With so much new data being collected from devices like that iWatch to Alexa on call to make every wish her command, this is only going to get tougher over time and more insightful.  Knowing what to do with the data collected is critical.

Key to governance is protection of consumer identity.  The data has to be protected.  You can create Disneyland for data but you have to be sure nobody gets hurt riding the “data” rides.

Second, you have to be clear about what patterns you are searching for in the data.  I always love those Eschler drawings where you can see several different images from the same piece.  Your perspective matters as you know what you see in it at first so clearly.  But with a bit of time you can start to see something new.

Finally once you find the patterns, you have to act.  Do you have an insight pulled from data?  Get going on exploiting that advantage.  These insights may not last so taking action is key.

-Ian

A Is For Analysis & I Is For Information

I prefer to think of AI as a tool that lets me analyze information rapidly.  It takes me out of the drudgery of digging for gold and into the “nugget finding” business.  I want those insights that allow me to recommend a new investment for Vestigo Ventures Fund I.

I know my way around SQL tools that let me build the right queries to help you as an investor in our Fund get returns.  It allows us to form a basis of fact about the industry.  We can believe we understand something but in reality using data discovered using our tools shows the facts.  We can then see if the facts are different than the conjecture.

This benefits our ability to find the right investments without human bias.  The tools allow us to dig into the right SEO and SEM strategies to build a business.  They also allow us to help a company in the portfolio defend their position as we see competitors emerge.  Al is key to helping a business growth.

Plus its just cool to see what the use of AI and our five petabyte database reveals to my inquiring mind.  Are robos any good at customer acquisition?  Answer is no.  It appears they are burning lots of funding without a return from the value of the assets raised.

Are online lenders good at digital marketing?  You bet they are quite good at taking traffic into their website and intelligently reselling some leads to others or finding ways to close a new relationship with novel methods like a meet up.  Why try this on our own using scarce resources when we can learn best practices on our own via our unique data set using our AI tools?

-Mike

LifeYield’s Tax Efficiency Score

Adding Talent-Harry Bartle Joins LifeYield

Netcapital On NESN

Blog-Customer Needs