Artificial intelligence (AI) is disrupting how companies operate. AI represents a foundational change in the way systems will be architected and delivered, and ultimately is a trend that is rapidly becoming a priority for Online’s clients and for Online.
Creating a foundation in artificial intelligence allows businesses to truly transform their partner ecosystems, business models, and customer experiences. Organizations that can uniquely combine architecture and technical innovation will play a major role in leap-frogging the competition, gaining true market advantage. But creating this foundation requires having the right skills at your disposal – and you need to get access to those skills now.
Over the last number of months, Online’s Innovation Lab has had numerous discussions with our clients to understand what trends they see as important, and which ones they
|
AI analyst reports make it clear that enterprises will be making substantial investments in finding specific use cases for AI, and exploring them more fully.
|
are wanting to invest in and explore. These conversations often include many of the expected topics - Blockchain, Artificial Intelligence, Internet of Things, Conversational User Interfaces, Virtual Assistants – as well as a few others.
AI is a foundational element and underpins all of the innovation themes we discussed; companies are not just interested but are investing in innovation around Artificial Intelligence (AI). The demand for AI and Data Literacy skills are increasing rapidly, and there is a sizeable short fall in the broad resource base to address this demand. Forecasts from major analysts indicate that a major portion of future application development initiatives will include AI, combining developers, and data scientists in creating adaptive solutions.
There is a lot of excitement around AI from Onliners and our clients. Let’s look at some of the areas where Online has invested in AI, including AI based systems that are currently in production, how AI will - and already is - deliver value to our clients, and future investigations we plan to make in the space.
Machine Learning and AI
Machine learning and AI are areas that depend upon large amounts of telemetry (information) for generating models of prediction and prescription.
There are many research papers available today that speak to how the work being done
|
AI is really about collecting large amounts of telemetry, applying smart analysis to the data (data scientists) and creating models to predict the key variables under study.
|
in AI can be applied to other areas such as buyer behaviour prediction. As an example, consider the systems today that monitor an end user’s browser, app, or other interaction behaviours, and provide analysis and prediction in real time. Think of “Other buyers of this product also purchased these products” examples you’ve likely encountered if you do any form of online shopping. Now ask yourself what went into their recommendations? How did they figure that out?
The major web players in this space have had access and continue to accumulate large amounts of telemetry. With this data, they can conduct large amounts of analysis and experimentation to create models which can be used to measure, predict, and alter the systems to ultimately drive results.
Online and AI Solutions: Introducing Monica
Online manages a wide variety of servers for our clients hosted on platforms like Sharepoint or Azure. These servers provide a lot of telemetry: everything from CPU and memory load, thru to Network, and whatever specialized service logging mechanisms exist.
Online also has Subject Matter Experts in the monitoring and management of these systems, as well as a limited (but very valuable) number of Data Scientists. Using these Subject Matter Experts, Online created Monica (Monitoring Information through Cognitive Application), an artificially intelligent assistant to monitor these services for our clients (click the image below for an expanded view of Monica).
Monica uses a modified LSTM (Long Short-Term Memory) Neural Network that can predict
the future state of server resource utilization (CPU, memory, and disk) with high accuracy. Data highlighting an anomaly can be immediately presented to IT staff using commonly available BI tools.
Monica intelligently creates a real-time baseline of the behaviour and load of the servers and services under management. Monica constantly makes very near term (seconds) predictions regarding what load a system will be under. If that load varies to a large amount from the specific model she has dynamically created for that system (days of week, hour of day, etc.), she will raise alerts.
She even has the nascent capability of making recommendations as to what actions could be taken to remediate the anticipated problem. Additional systems can be easily added to Monica for her to monitor.
Also, Monica does not sleep.
The Creation of Monica
Monica went into production in May, 2017 and is the brainchild of Onliner Jeff Adkin, with support from Onliner Paul Hagen, along with some suggestions from me. Monica is a powerful AI assistant for our Managed Services teams at Online.
Many innovative machine learning and AI experiments were conducted in the creation of Monica. The Innovation Lab will be publishing a paper, “Real-time Prediction of Infrastructure Issues using Machine Learning” very shortly which describes in greater depth the problem area that Monica addresses, and even the math and analytics associated.
Online and AI in the Future
Online works with many clients in the IOT space. In IOT, there is a lot of telemetry, a lot of “mini” servers in the form of either controllers or devices with more intelligence requirements, and very large distribution of things. The system itself is cloud connected and distributed, but the controllers have limitations on CPU, Memory, Bandwidth, and other components. Systems and approaches such as Monica can aid greatly in creating a system that is reliable, adaptable, and sustainable. Online is working on adapting many of the techniques of Monica to address the needs of this system. AI will play a major part in being successful in this endeavour.
What you can do today
I often get asked by clients or Onliners about how they can improve their understanding of AI and get a feel for what the data science process looks like. Here are a few suggestions.
- Look for some courses on Udemy and Coursera. I have found a plethora of either introductory or advanced usage in Machine Learning.
- There are also many different Machine Learning and Data Science tools out there.
- I have been looking at KNIME for a short amount of time (very powerful), and have focused more on H2O (H2O.ai). H2O is Open Source and free to use / work with (although their Driverless AI is licensed).
- The following video highlights some interesting areas in Data Science with H2O: https://youtu.be/3_gm00kBwEw
- Most importantly, the leaders at Online for AI would recommend getting your hands dirty in the guts of data preparation, coding the algorithms themselves, and understanding what is going on under the hood.
AI is a foundational area for IT. I am excited by the initiatives, investments, and actual production benefits that Online has received from our own investments in AI, and strongly believe that Onliner’s can contribute to our client’s success with their initiatives. Please be on the lookout for invitations to participate in the Online Innovation Labs process as we move forward.
Is your company looking into starting to use AI? Maybe you already have some projects on the go that are utilizing AI in some form? I'd love to hear from you! Feel free to leave a comment below or email me directly.
Submit a Comment