AI


There is no one way to ‘deploy AI’ One fundamental problem when discussing AI use by MSPs—


and indeed by any organisation—is that there is no single way of implementing this technology. Tere are multiple approaches, and these will, over time, likely be used in combination. In fact, AI adoption falls into four distinct categories, each with its own complexity and resource demands: Use, Tune, Combine, and Build Your Own (BYO). Te simplest implementation is the use of AI in vendor


products. MSPs will be adopting products with integrated AI and will either be using these as part of their everyday service provision, or actively selling this as an upgrade to their customers. With these ready-made tools, you can leverage sophisticated AI capabilities—whether it’s improving customer interactions or streamlining internal processes—with minimal effort and technical expertise. It’s an efficient, low-risk way to start integrating AI, especially if you have little to no proprietary data. For MSPs, this might mean adopting AI-powered security services that improve threat detection or using AI-enhanced backup solutions to minimise false alerts. If you want to customise AI a bit further, the Tune approach


lets you adjust existing AI models with your own data. Tis is particularly useful for applications like generative AI, where personalisation makes a significant difference. For example, you could train a chatbot on your company’s customer service logs, making its responses more relevant and accurate. By taking a set of records and augmenting a Large Language Model (LLM), through a process known as Retrieval Augmented Generation (RAG), an MSP can create generative AI output that can be used as part of their processes. And then this can go further, with MSPs creating their own large action models, an artificial intelligence model that can understand and execute complex tasks. For businesses with more complex needs, combining multiple


AI models or components can be a powerful strategy. Tis approach is perfect for those who need a more flexible, integrated system without starting from the ground up. Platforms like Vertex, Bedrock, or Azure AI Studio allow you to piece together different AI models to create a more comprehensive solution. For MSPs, this could mean combining AI models for security, data analytics, and automation. Tis method works best if you have thousands of high-quality records to train and validate these composite models. At the highest level of complexity is building your own AI


models. Tis approach is for organisations that require solutions tailored to their specific needs and can achieve the ROI to justify the significant resources—vast amounts of data, advanced technical expertise, and a considerable financial investment. Typically, you’d need hundreds of thousands of records to train a reliable model. Tis is the level for companies that have unique, data-rich environments where a proprietary AI system could give them a competitive edge. MSPs at this stage may be developing their own AI-based solutions for automation, cybersecurity, or data management, offering services that can’t be easily replicated. Tere is a huge difference in the time and investment involved


in these different approaches. Tey all raise different questions that the MSP would need to address—can we trust the vendor’s AI technology? Do we see the use of off-the-shelf AI technology as a


www.pcr-online.biz


stepping stone towards creating our own models? Should we be expanding our teams to build differentiation through AI models ourselves? And how can we make sure our customers’ data is safe and they are protected at all times?


Understand the why and ROI Regardless of which level you choose, understanding the rationale and return on investment (ROI) is critical. AI can’t be implemented just for the sake of it—it needs to serve a clear purpose. For instance, AI-driven customer support offers tangible, measurable benefits: if AI helps close five additional tickets every two hours without human intervention, that’s a clear, quantifiable advantage. Tere is a great deal of fear around lost jobs as AI improves.


But fundamental to the adoption of AI will not just be in the creation and adoption, but in understanding its output and its flaws. AI, when done poorly, is a ‘black box’, which makes it difficult to understand what’s happening and when it goes wrong. We need people who understand these flaws to tweak the inputs and improve the output. AI’s real power lies in augmenting human capabilities—not replacing them. MSPs that embrace AI with a clear strategy, understanding


of their data, and the right level of deployment will be better positioned to thrive in this evolving landscape. Pillars for success With ‘implementing AI’ meaning so many different things,


and a constant need to question why a business is doing it at all, it would be easy to lose sight of crucial issues such as customer safety and data protection. By basing any AI implementation, complex or simple, on some


fundamental pillars, these issues become easier to bake into any project. Any question can be answered by thinking about these pillars, and how they relate to the matter in hand. A business should develop their own with reference to local laws and their customer base, but these are a good place to start: Consent: Opt-in is our fundamental principle, and we ask for


your consent before you engage with any AI in our products or services. Affirmative consent ensures we respect user preferences and privacy while mitigating risks associated with unauthorised data usage. • Transparency: We prioritise accountability, education, and informed decision-making; and we’ll tell you when and how we’re using AI tools with your data. Te more knowledge our partners have, the more confidence they’ll have to explore the possibilities of AI. • Context: We strive to deliver maximum value and effectiveness through careful, efficient use of technology. Tis means we’ll continuously work to understand the unique challenges and requirements of each situation and narrowly tailor AI solutions to each use case. AI is not just an opportunity to create efficiencies and improve


how an MSP works, but also an opportunity to sell AI hosting and language models to customers. But only by building AI products, whether created internally or bought from vendors, on pillars that keep customer data safe and your own reputation intact, will an MSP be a credible vendor for these technologies.


September/October 2024 | 41


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52