4 Reasons Why Investment Banks are Choosing To Buy NLP Instead of Build

Tejas Shastry

September 12, 2020

Blog

Natural Language Processing (NLP) has become essential for many institutions as they struggle to keep up with the immense amount of voice, chat, and email conversations across their firm.

For many institutions, NLP is transformer that converts unstructured data (electricity) into the proper voltage to power applications like chatbots, insight reports, conduct surveillance, and more.

As investment banks shift to remote and distributed work, many are looking at leveraging NLP face a common dilemma: build vs buy.

Institutions are motivated to build internally to protect IP and their competitive edge and often wonder if off-the-shelf NLP solutions will really fit their workflow.

Here at GreenKey, we’ve recently seen the dial tilt completely the other way to buy over build, and here’s the top 4 reasons why.

1) Better return on investment

Hiring an internal team of data scientists is expensive, not to mention the hardware and infrastructure required to train models on large data sets. When it comes to many projects involving NLP, firms are increasingly seeing that they can buy for much cheaper than they can build.

We ran the numbers. Buying a natural language processing platform with pre-trained models and customizing it for a particular use case can cost as little as 5% of what it would cost to build it yourself.

Think of buying GreenKey as paying for a single employee as opposed to a team of 20 to build it yourself.

2) Focus on alpha

NLP is a broad field encompassing all areas of the ability of machines to understand and produce human language. Within NLP, there are many different “tasks” that a machine can perform.

Vendors like GreenKey have focused on building models that provide many of the essential tasks out of the box, such as recognizing a client inquiry and extracting product details.

Buying an NLP platform to get up and running enables investment banks to focus their data science attention on higher level integrations that can truly generate alpha against competitors.

3) Avoid sunk-cost thinking and leverage state of the art

Some firms have invested so much money into building their own NLP platform they can’t make the decision to leverage a vendor.

Let’s face it – you can’t build good NLP in a vacuum.  You want to leverage best in breed technology and ideas, and only a vendor working with the community will expose you to that.

The field of NLP is moving at lightning speed, and it’s often hard to keep up. At GreenKey, we use state of the art transformer-based NLP models that can be customized from a handful of examples.

By buying pre-trained models, investment banks get to leverage both state-of-the-art architectures and also a large training corpus that extends far beyond the data they have in house.

4) Get to production faster

Time is often more valuable than money for many banks, and adopting an NLP platform built internally can take months or years. Buying an NLP platform that helps power a bank’s own efforts in automation, building chat bots, and understanding client relationships can help greatly accelerate time to market.

The faster banks can understand their data, the faster they can make more money, compounding the ROI of buying over building.

At GreenKey, we offer additional help through our Professionals Services package to help banks deploy NLP across their institutions and integrate directly with their own data sources.

 

Increasingly the dilemma of build vs buy when it comes to NLP seems to be resolving: buy an NLP platform to get you started so you can focus your attention on the areas that matter most.

Interested in GreenKey as your NLP partner? Contact us today to get a demo.