Natural Language Interface In the Enterprise

DZone 's Guide to

Natural Language Interface In the Enterprise

Let's take a look at why enterprise application companies should take a cue from Apple's Siri and Google Assistant.

· AI Zone ·
Free Resource

Why Enterprise Application Companies Should Take a Cue From Apple's Siri and Google Assistant

Enterprise applications are the next frontier in the adoption of natural language interfaces. Unlike consumer tech, e-commerce, and various chatbots where NLP/U is more of a technical novelty, the world of enterprise is a killer ground for natural language interfaces.

A Need for a Unified Interface

One of the key unique properties of natural language is the fact that it provides a unified interface to any data source or sources. It's the one interface that everyone already knows, and at the same time, it's the same interface to any supporting system. Think about it...you can easily ask a lawyer, salesman, or marketing professional about any specific topic as long as you can formulate a question in a minimally understandable way.

To grasp this immensely powerful fact, let's look at an example. Here is a simple ad-hoc question that could just have easily popped up in any marketing or sales meeting in your company:

Is there a correlation between sales in NY state for the last 2 months and the AdWord campaign we've run in the same period and the same region?

To understand the massive inefficiencies behind today's silo-ed data analytics systems and associated business processes, let's deconstruct and see how an organization could answer this very question today.

Current Inefficiencies

Despite the perceived simplicity of this question, the tedious process of finding the answer to it is anything but simple or efficient.

First off, the data required to answer this question resides in two vastly different systems that are likely installed, maintained, and managed by two different groups within your company:

  • Google Analytics (or Marketo or HubSpot) where a company manages its AdWords campaign and generally collects web analytics data, and
  • Salesforce.com (or other sales CRM) where sales data gets collected

Because these systems are vastly different and incompatible (different user accounts, different interfaces, different data model, different query languages, etc.), it is very typical that they are managed by different people from different groups within a company.

Now, let's deconstruct the process of answering our question in a typical small- to mid-sized (SMB) company:

  • Since it will involve two different groups to work together, a project manager should be assigned (often a business or data analyst, or someone from the reporting department).
  • The project manager (or the entire group) should spec out what a desired correlation really means (i.e. the essence of the question); it's likely that the number of AdWords clicks and bookings on a given date should be a strong indicator of correlation. For simplicity, we'll leave out filtering out other influencing factors such as holidays, promotions, other marketing programs' cross-pollination, etc.
  • Google Analytics (GA) personnel should now develop and run the report in the GA user interface for the given time range that would include the number of clicks per date for specified AdWords campaign and in a specified geographical region.
  • Once the GA report is complete, it should be exported into CSV format and sent over to the Project Manager.
  • The Safesforce.com (SFDC) person should create (using SQL-like query language) and run a report that shows bookings per date for the given date range and in a specified geographical region.
  • Once the SFDC report is complete, it should be exported into the same CSV format and sent over to the Project Manager.
  • Once the Project Manager receives both reports, she would combine them into one by importing both CSVs into SAS, Excel, or other reporting tools and de-dup, clean up (date formats can be different, ensure that geo-regions are uniformly named, etc) and add correlation metric calculation as well as representative charts to visually show the correlation (or lack of thereof).
  • At the end, the resulting data tables and charts are exported into PDF or PowerPoint and sent back to the people that had the original question.

Depending on the agility of your company, this (unfortunately typical) boondoggle can take hours, days, or often weeks to complete.

All just to answer a benign ad-hoc question that popped up in the meeting...

This example illustrates a simple fact: companies found ways to effectively collect and store data but are woefully inefficient in allowing thousands of business users in their companies to access and explore that data. This is also the reason behind the fact that the vast majority of business users don't ask questions and therefore don't rely on data in their decision making: who in their right mind would trigger such a costly process to answer just a simple question?!

In fact, the only people that have efficient access to that data are the people who don't have any actual questions — your business and data analysts who only exist to find answers to other people's questions.

Natural Language Interface

When you read the tedious and cumbersome process above, remember that this whole thing started with a simple question:

Is there a correlation between sales in NY state for the last 2 months and the AdWord campaign we've run in the same period and the same region?

It's funny that anyone can simply say or type this question without any need-to-know specifics of Google Analytics, Salesforce, spreadsheets, etc. No project managers, no data cleaning, or aggregation. Nothing but a basic ability to formulate your sentence.

It is also interesting to note that mere 35-40 years ago, searching the nascent internet was an engineering task: you had to know Perl or CGI scripting or Shell and write a pretty complex program to find something on the net. Well, today you just type or speak to Google and get your answers.

Apple's Siri is the case in study of how natural language is becoming a unified interface to the entire hardware and software ecosystem for Apple. In a short 7 years, Siri has transformed from a cute app on the iPhone 4s to a unified standard interface among iPhone, Watch, HomePod, TV, and MacBooks, and it's rapidly advancing with every OS release. The same is happening with Google Assistant and Google's portfolio of hardware and software.


Enterprise software companies would be wise to take a cue from Apple and Google on that. The crossover from consumer to enterprise happens quicker and quicker and enterprise applications companies can one day find themselves behind and outdated compared to new AI-powered newcomers. Based on what we see and hear at DataLingvo, this will likely happen in the next 24-36 months.

artificial intelligence, enterprise application companies, google assistant, machine learning, natural language processing, nlp, siri, unified interface

Published at DZone with permission of Aaron Radzinski , DZone MVB. See the original article here.

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}