Founded in 2005, Rimini Street is a provider of support services for enterprise applications from companies like SAP, Oracle and Salesforce.com. The goal is to reach $1 billion in revenues by 2026.
And to accomplish this, the company evaluated a myriad of AI solutions. But they all fell short.
“In the end, we leveraged a combination of best of breed open source technology and tools to build our own AI platform that is very powerful, intelligent, extensible and scalable,” said Brian Slepko, who is the Executive Vice President of Global Service Delivery at Rimini Street. “Our home-grown AI platform and components give us all the flexibility we need along with the data intelligence–and we now have three patents pending as a result.”
It was a risky strategy. After all, many AI projects fail to go beyond the proof-of-concept stage.
But to remain competitive, companies often have little choice but to make big bets with AI. “Creating your own AI solution will require a huge budget, research, time, and resource commitment,” said Swapnil Bhagwat, who is the Vice President of Marketing at Atlas Systems. “But AI capabilities for any business have the potential to become a great competitive advantage.”
So then what are the factors to take into account when looking at whether to build AI?
Let’s take a look:
Is AI The Really Needed? Yes, the temptation is to assume that AI is some kind of elixir for any type of problem. But this is a big mistake. True, AI is very powerful but a simpler technology may be a better approach.
“Identify your business challenges first–don’t jump on the AI/ML bandwagon because it is the topic du jour,” said Slepko. “Put together a very structured plan which includes your vision, strategy, how you want to use the technology and the desired business outcomes you are trying to tackle.”
Experience: If your organization has little real-world experience with AI, then you might want to first look at off-the-shelf products. For example, a chatbot could be a good place to start, such as with basic customer support services.
Another idea is to look at your company’s IT applications and apply the AI functions. They can be quite powerful, especially since they are already integrated with your data sources.
Don’t Do Everything: If you decide to build a home-grown system, it is still smart to look at ways to get help from other companies. An example would be to outsource the data labelling (the process can be tricky and laborious).
Capabilities: Is your organization AI ready? Usually the answer is “no.”
According to Mingkuan Liu, who is the Senior Director of Data Science at Appen, here are some questions to ask:
- Is AI critical to your business or project?
- What is the investment level for this project?
- What is the visibility and participation of executives?
- What is the focus on bias, risk, governance, and ethics?
Note that Appen does have a useful free AI Readiness tool to help out.
Data: This is often underestimated. But data is absolutely essential for AI success.
“It’s important to know that generic AI platforms require a massive data integration effort and an army of data scientists to implement it,” said Bill Scudder, who is the Vice President and General Manager of AIoT Solutions at AspenTech. “Organizations should consider finding solutions that can seamlessly integrate data sources and deliver actionable insights. This allows hundreds of cases to be run in minutes to determine the best ways to increase margin.”
