faster iterations and, ultimately, highervalue outputs. But the mainstream market wants results the first time, and that’s something MSPs can help them get in a way they wouldn’t be able to guarantee on their own. As you advise your clients on AI, seeing provable business results depends on careful consideration across four core pillars. Data and Use Cases If there’s one lesson every AI initiative has reinforced so far, it’s the need for clarity. Artificial intelligence tools are educated curators and creative regurgitators, meaning their usefulness is limited by the data that feeds them. MSPs have been beating that drum for what feels like years: high-quality, wellstructured data is non-negotiable. But clarity doesn’t end with the inputs. Equally critical is defining the output. What’s the use case? What business outcome do they want AI to help achieve? Too often, clients start with “we need AI” rather than “we need AI to do this.” MSPs can help bridge that gap, guiding clients to articulate both where they’re starting (the data) and where they’re headed (the outcome). That clear line of sight from inputs to results is what separates experiments from projects built for success. Strategy Midmarket organizations considering any project must think strategically and act tactically, even when starting with pilots. They don’t have the budget to do otherwise. Once the inputs and outcomes are defined, the real work begins: defining the path between them. Every AI rollout involves dozens of choices – from selecting the right models and tools to establishing governance guardrails to planning for scale. Each decision shapes the trajectory and increases the risk of veering off course, and midmarket organizations rarely have the time or budget to reinvent the wheel. MSPs give them a scalability advantage: enterpriselevel lessons without enterprise-level trial-and-error. With managed AI services and a portfolio of projects to draw from, MSPs bring the benefit of hindsight: spotting patterns, avoiding common pitfalls and tailoring proven approaches to fit each client’s goals. That perspective helps clients move with confidence, knowing their investment is backed by a structured plan for return. Execution Even the best strategy falls flat without effective execution. An investment in AI only delivers value if teams adopt it – and not just adopt it but use it well. That means more than standing up a model: it takes training, clear implementation procedures, seamless integration into existing tools and workflows, and incremental rollouts that build confidence without disrupting business. Above all, execution must leave employees with a faster, more effective way of working than what came before, or the project will be effectively dead before it’s off the ground. For many organizations, this is where the real hurdles appear. Securing adoption, avoiding “rip and replace” panic, and ensuring early wins demand a level of expertise these midmarket organizations don’t yet have. MSPs can step in here as a guide and safety net, offering proven playbooks, surfacing challenges companies might not anticipate and making sure execution stays aligned to business goals. This is also where MSP can create continuity – keeping the rollout on track day-today and ensuring adoption doesn’t stall after the initial push. Measurement & Recalibration No AI initiative is ever “set it and forget it.” Models drift, business goals evolve and inputs lose clarity over time. Even the best rollout will require ongoing adjustment. The key is to treat AI as a living system that needs to be measured, reviewed and refined over time – not as a one-time deployment. Built-in reporting and regular reviews ensure AI delivers real ROI and stays aligned with business objectives. But the challenge for most organizations is knowing which numbers matter. Chasing vanity metrics can give the illusion of progress while masking real issues. MSPs can bring discipline to this process. With both technical insight and business context, they help clients measure what matters, identify when recalibration is needed and course-correct before small problems become costly setbacks. In doing so, they turn AI from a fragile experiment into a sustainable, continually improving capability. Early adopters may have been willing to experiment, but midmarket organizations can’t afford that luxury. By guiding clients through data, strategy, execution and recalibration, MSPs can turn AI from a costly experiment into a practical engine for measurable business value. o Michael Gray is CTO of Thrive Technology Adoption Life Cycle Source: B2U Agents enabled by generative AI could function as hyperefficient virtual coworkers Source: McKinsey & Co. Illustration of how an agent system might execute a workflow, from prompt to output Using natural language, the user prompts the generative AI agent system to complet a task. The agent system interprets the prompt and builds a work plan. A manager agent subdivides the project into tasks assigned to specialist agents; they gather and analyze data from multiple sources and collaborate with one another to execute their individual missions. The agent team shares the draft output with the user. The agent team receives user feedback, then iterates and refines output accordingly. External systems: Agents interact with databases and systems– both organizational and external data–to complete the task. Manager agent Specialist agents Start End Analyst agent Checker agent Planner agent 1 2 3 4 The steep drop from pilots to production for task-specific GenAI tools reveals the GenAI divide Source: MIT Project NANDA Why GenAI pilots fall: top barriers to scaling AI in the enterprise Users were asked to rate each issue on a scale of 1-10 Challenging change management Lack of executive sponsorship EARLY MARKET INNOVATORS 2.5% TECH ENTHUSIASTS EARLY ADOPTERS 13.5% EARLY MAJORITY 34% LATE MAJORITY 34% LAGGARDS 16% THE CHASM MAINSTREAM MARKET VISIONARIES PRAGMATISTS CONSERVATIVES SKEPTICS General-Purpose LLMs 80% 60% 50% 20% 40% 5% Investigated Piloted Successfully Implemented Embedded or Task-Specific GenAI 14 CHANNELVISION | FALL 2025
RkJQdWJsaXNoZXIy NTg4Njc=