Editorial
Everybody wants the efficiency gains from AI, and they want them yesterday. But how exactly do you successfully implement AI in an organization? When I asked an AI platform that question, the response was spot on: “You need to prove it works before you trust it works.”
Proving the benefits lets you create scalable, trusted models that fit within a complex IT infrastructure. In my experience, three key factors lead to AI acceptance: a clear scope, high-quality data and well-defined ownership. In this article, we’ll explore each of these issues and understand how they contribute to AI's long-term success.
1. Define a Tight Scope for AI Implementation
Even with technology as promising and transformational as AI, you can’t jump straight into wholesale adoption. Instead, you need a process and strategy that is intentional, measurable and aligned with your operational goals.
Start with areas where you have a lot of structured, process-oriented tasks. When you test AI with repeatable, high-volume workflows, you can evaluate its utility at each step and build feedback loops for incremental improvement. We typically recommend starting with these three departments:
- IT: You’d be hard-pressed to find a better use case for AI than high-volume help desk tickets. AI is ideal for classifying and summarizing tickets, sending auto-responses and providing predictive analytics — all of which help reduce mean time to resolution (MTTR). That’s why many IT departments now use AI assistants to handle password reset requests and automate other repetitive tasks. You can also combine AI with other technology for even more eye-opening results. We worked with one financial services firm that used sensor-triggered automations to prevent 207,000 IT tickets monthly.
- Finance: As Gartner noted, nearly 60% of finance functions used AI last year, which makes sense, given AI’s strength in handling structured, rules-based workflows. Yet there’s still much more room for growth, with finance “primed for a generative AI revolution,” according to IBM. With vast amounts of financial data, there’s potential for significant ROI, whether you’re using a Microsoft AI agent to automate invoice processing and save “millions of dollars in shipping costs,” or gaining 20% cost savings by using AI-powered software within your financial operations model, as IBM reported.
- HR: “AI… can probably do 50-75% of the work we do in HR.” That’s a bold proclamation from an HR consultant, but consider what we’re seeing in the field. For example, Chipotle used an AI assistant to cut the time for a job applicant to start working “from 12 days to four days.” The value of AI isn’t simply in handling predictable tasks; it also allows your employees to focus on strategic, higher-value initiatives so that you can tap into the “unrealized human value” throughout your organization.
Related Article: Is Your Data Good Enough to Power AI Agents?
2. Fuel AI With High-Quality Data
Despite its many well-documented benefits, “AI doesn’t know good data from bad,” as a recent Enterprise Strategy Group report noted. “It just knows data, and whatever data is being used to train a model will ultimately become the model.” That’s one reason why the number-one challenge for CIOs is often getting access to quality data.
Obtaining this high-quality data means you must be intentional about data characteristics. Is your data accurate? How frequently is it collected? Do you have endpoints and sensors monitoring your laptops and other devices 24/7? Data is the foundation for AI, and your results will only be as good as your inputs.
AI also requires structured data to maximize its value, which is why more than 70% of top finance organizations have standardized data architecture, as IBM noted. This data structure allows companies to do more with AI, including defining underlying KPIs and driving results. As I noted in a previous article, with the rise of agentic AI (which relies even less on human interaction), you need to pay even more attention to the data that’s fueling these agents.
3. Establish Clear Operational Ownership of AI
As with any tech platform, IT should take the lead in building trust and implementing AI throughout your organization, especially since testing and deploying AI leverages the strengths, skills and experience that IT brings to the table.
For example, AI can expose your organization to risk, partly because it’s not always 100% accurate. If a VP places a $2 million order based on an incorrect AI-driven sales forecast, that’s a costly mistake. But the risks don’t end there. Just as employees bring their own devices to work, Microsoft reports that 78% of AI users bring their own AI tools to work (termed BYOAI), which is a significant concern if users are uploading and processing company data. IT can educate employees on best practices — just remember that IT may have biases and blind spots.
IT should also collaborate with teams to manage specific AI tools. From onboarding and governance to managing the ethics of AI, everything starts with IT, which is why it’s crucial to keep upskilling your IT staff.
The bottom line is that a strategic approach to AI implementation needs to be built on trust. As that Enterprise Strategy Group report noted, “Without trust, full adoption can’t happen. And without full adoption, the overall goals can’t be achieved.” But given AI's long-lasting impact, making that investment now and proving AI’s effectiveness with short-term wins will likely pay off with significant future returns.
Learn how you can join our contributor community.
