Businesses know that machine learning systems and artificial intelligence programs can be customized to meet a company’s specific needs. Most are at a loss to know how to begin developing them.  Many are worried about teaching a machine learning system their pivotal secrets and losing rights to the system itself.

We have talked and written extensively on the risks of artificial intelligence that many business executives may be too intimidated to seek its rewards. This hesitation is unwise and unnecessary.

Probably the best way to resolve these concerns is for companies to start building their own AI for improving internal processes.  While the prospect may seem foreign and scary, a universe of open source tools exists to make it easier. Running to GitHub and grabbing the best tools can make the entire process manageable and using open source in this manner can assure that a company owns what it creates.

In a previous blog post I posed my theory of “mystery equals margin,” suggesting that keeping the value of a product or service secret allows the provider to charge more than it could under perfect market information. AI development qualifies for the same aphorism, because, for non-technical business people, such uncertainty clouds the nature of creating a machine learning product that developers can bamboozle their customers into financial and contract terms deeply favorable to the vendor.

There are ways to strip the hocus pocus from vendor patter about how to create the tools your company needs.  One easy way is to hire a consultant who understands that process under an agreement that this consultant will not design the product, only check up on the processes, procedures and prices of the ultimate development vendor when selected.  This consultant may also assist with validation and testing after the AI programs are delivered.

Another way is to assign an internal software designer or development team to teach themselves machine learning and report back to the business executives. This route is getting easier month after month as free tools and tutorials gain and grow online. There are even new tools for the executives themselves to learn the process and stay ahead of the game.

For example, Uber Ludwig is a tool for non-programmers to both understand machine learning and to develop their own systems without the need for learning programming languages. Ludwig is a toolbox built on Google’s well-known open source machine learning platform TensorFlow.  In other words, it will show you how to train and test deep learning models without needing to write code. All you need is the appropriately formatted data file, a list of columns for inputs and a second list of columns for outputs, and Ludwig will train the AI program for you. This is also a great tool for developers to build complex models quickly and test their mettle before building out the final code.

The orange platform is based on a graphical user interface, allowing non-coders to “mine data, crunch numbers and derive insights.” Orange consolidates all of the functions for data manipulation, transformation and data mining so that an executive can see the entire process as a single workflow, including sophisticated visualizations and graphing of data sets. Orange is becoming a popular teaching tool for professors to demonstrate statistical insights and number crunching.

Of course, you can pay for well-supported company-managed AI service tools like Data Robot and Azure Studio if you prefer a curated experience. You can use any of these tools to begin understanding how to train an AI for your company, but if you know what your company needs, then look for tools that match your needs.  For example, some tools are excellent for unsupervised learning, providing groupings and determining deviations from your data sets.  Others are better for supervised learning, feeding both preliminary sets and result sets to provide predictive analysis. Quick reviews online can show you what you need.

Machine learning is a new type of program development, but it doesn’t need to intimidate your company. Finding a Sherpa to walk you through the mountain passes may be the easiest way to learn how this new technology can help your company.  But the right tools exist to teach yourself.  It may be time to start learning, so that your business can benefit from the machine learning explosion before your competitors use AI to eat your lunch.