How Microsoft MSPs Can Build Their AI Value Proposition

In a Sept. 18, 2024, post in the LinkedIn Official Blog, SVP and General Counsel Blake Lawit wrote:

In our Privacy Policy, we have added language to clarify how we use the information you share with us to develop the products and services of LinkedIn and its affiliates, including by training AI models used for content generation ("generative AI") and through security and safety measures.

When it comes to using members' data for generative AI training, we offer an opt-out setting. At this time, we are not enabling training for generative AI on member data from the European Economic Area, Switzerland, and the United Kingdom, and will not provide the setting to members in those regions until further notice.

As technology and our business evolves, and the world of work changes, we remain committed to providing clarity about our practices and keeping you in control of the information you entrust with us.

European readers will recognize the scent of GDPR in the stricter rules imposed in Europe, but it's very important to recognize why LinkedIn is making an announcement like this.

It's All About the Content, and Who Owns the Content
In a previous blog, I talked about the need for MSPs to align themselves with legal counsel and resources that can help them help their clients avoid trademark infringement when training large language models (LLMs) for use in AI applications. This change in LinkedIn's privacy policy to better manage member content and how it is used for generative AI purposes signals yet another proactive, deliberate move by Microsoft, its owner, to protect content rights.

In the earliest days of generative AI (which seem like just yesterday), LLMs were routinely trained using large masses of content drawn from the Internet. It quickly became obvious that much of this content was copyrighted, trademarked or otherwise protected. It also quickly became obvious that it would be incredibly difficult to detect and identify protected source content used by examining the results output by the AI engine.

At the same time, those outputs seemed somewhat limited. All media coverage focused on how AI could write e-mails, reports and other written content. It could also take photographs and make you look more rugged, more desirable, more alien, whatever you liked. Practical, valuable business applications were scarce at best.

AI Business Applications Are All About the Data
"Understanding where the data is, how to get access to it, how to harness it, what [and] who owns it -- all that fun stuff -- is crucial to the success of anything AI," explains David Tan, CTO at AI developer CrushBank.

Tan bristles when asked what MSPs are doing with AI engines like Microsoft Copilot. "Copilot will streamline some of your daily tasks. It's one thing to say, 'Yes, you should use Copilot to write or make a deck,'" says Tan. "What everyone doesn't understand is that even in a best-case scenario, the data that lives in your Microsoft 365 tenant is half the picture, probably way less than that."

He continues, "How are you building a business-impactful solution with only half the data?"

The suggestion that MSPs and their customers are only using a fraction of what Copilot can actually do is not unfamiliar. The same has been said forever about Excel: It's a very powerful spreadsheet, but most users only use about 10 percent of what it's capable of. With generative AI like Copilot, however, there is a strategy for enabling customers to enjoy a much larger proportion of its real power, an opportunity Tan suggests MSPs are currently leaving on the table.

If It's All About the Data, MSPs Need to Know the Data
The key for MSPs who want to build a successful generative AI practice is to learn how to truly know their customers' data. This requires hiring, or training to become, a data analyst. Data analysts gather, process and interpret data to help businesses solve problems and make informed decisions. They may collect data through surveys, tracking Web site visitors or purchasing datasets. They then assure the quality of data by removing errors and duplicates.

Data analysts also create database structures, deciding what data to store and where, how to categorize data and manage it, and how the data will be made available. Once the data is properly analyzed and organized, it can be used to train LLMs that can then be used to perform all manner of tasks, or augment the accuracy and depth of understanding available from the results.

Way back when, as a Cisco reseller, I learned all about routing and switching. In fact, most of the training that MSPs receive today comes from manufacturers or learning centers authorized to train for the manufacturers. Generative AI will be no different. The companies who are today developing or integrating AI into their platforms are performing the required data analysis and LLM training themselves. They are the only ones who currently possess the requisite skills.

But this is a business model that cannot possibly scale. For those companies to grow their markets and broaden their success, they must do what every successful manufacturer or developer of anything has done for over forty years: They must turn to our channel.

The good news for you, the evolving MSP, is that when they turn to you, they will assess your current level of capability and train you to a much higher level where you can enjoy the ability to build truly valuable, practical applications for your customers. When you are doing this, you will find yourself also enjoying a far closer, more intimate relationship with those customers as it will be your work that powers their success. Time to seek out the AI partners of your choice.

About the Author

Technologist, creator of compelling content, and senior "resultant" Howard M. Cohen has been in the information technology industry for more than four decades. He has held senior executive positions in many of the top channel partner organizations and he currently writes for and about IT and the IT channel.

Featured