How to Plan Your AI Strategy

In our last discussion, we established the foundational concepts of AI—what it is, its core capabilities, and the risks leaders must understand. Now, we move from understanding to action. If you haven’t read the foundations post yet, you can find it here.

The most common failure mode I see in AI adoption is the leap from initial excitement directly to a series of random, siloed pilots. AI is a tool for scaling and optimization. You cannot optimize what you do not understand. And, you should not scale what you don’t understand. Unfortunately, I often see AI being used to do things the teams don’t know how to do well or barely understand; this is a recipe for wasted resources and disillusionment. A successful AI transformation is not an impulse-based try-out; it is an act of architecture. It requires a blueprint.

The big question is, how do you create that blueprint? How do you move from a vague desire to "use AI" to a disciplined, quantified, and unified strategic plan?

Based on my two decades of experience helping companies transform their operations, here is the three-phase process I use to build that plan.

Phase 1: The Foundation — Create a Workflow Catalog

The first step is to create a comprehensive catalog of your key business workflows. You can start with a single team or department to build momentum. The goal is to document the "as-is" state of your operations. I admit this is going to be boring, and it’s going to be hard. This is the early morning workout that makes star athletes who they are.

For each workflow, identify the key tasks, their sequence, the resources and tools involved, the cost of those resources, and the toil. Toil is a gut check from the operators of the workflow that indicates how painful the process is on a scale of 1 to 3.

But here’s the crucial part, the detail that transforms this from a simple documentation exercise into a strategic one: classify each task into one of three categories: Operate, Create, or Innovate.

• Operate tasks are the engine room of your business. They are repetitive, process-driven, and essential for keeping the lights on. Consider tasks like compiling weekly reports, performing data entry, or logging information into Salesforce.
• Create tasks require a degree of originality but still follow established patterns. This includes drafting a blog post from an outline or writing a personalized sales email.
• Innovate tasks are where true competitive advantage is born. They require imagination, out-of-the-box thinking, and deep strategic work. This is coming up with a disruptive new feature idea or designing a new go-to-market strategy.

When you complete this exercise, you will likely notice a pattern: your team's time is disproportionately spent on "Operate" tasks, the very tasks that create the least amount of enterprise value. The strategic goal of your AI initiative is to invert this pyramid—to use AI to automate "Operate" work, augment "Create" work, and liberate your most valuable human talent to focus on "Innovate."

Phase 2: The Path — The Automation Ladder

Once you have a catalog of what your team does, the next question is how AI can help. To answer this, I use a framework I call the Automation Ladder, which classifies any potential AI automation into one of three levels: L1, L2, or L3.

L1: AI-Assisted (Direct Task Execution)
These are one-step tasks where a human provides a direct input and receives an immediate output. This is about using AI to assist you with something you are already doing. Think of using a prompt template to draft a blog post from an outline you provided. You are in complete control of the input and the output at all times.

L2: AI-Augmented (Instruction-Following Automation)
These are multi-step automations that follow a clear, predefined sequence of instructions. Imagine compiling a monthly competitor research report. Your process would be: 1) get a list of competitors from the CRM, 2) browse the web for their product announcements, 3) synthesize the findings, 4) draft the report, and 5) email it to the team. An L2 automation uses a combination of AI and tool invocations to execute this exact sequence. It is following your instructions, just at machine-scale.

L3: AI-Autonomous (Goal-Pursuing Automation)
This is the most advanced level. Here, you do not provide instructions; you provide a goal. Imagine giving an agent the goal of "getting 500 trial sign-ups for our new product." The L3 agent, with access to a suite of tools and sub-agents (like an email copywriter, an ad manager, etc.), would then devise, orchestrate, and optimize its own plan to achieve that goal. It is pursuing an outcome, not just following a script.

By mapping each of your cataloged workflows to one of these three levels, you are no longer just looking at a list of processes; you are looking at a clear map of your company's automation potential.

Phase 3: The Filter — The Prioritization Engine

You now have a map. The final step is to decide where to go first. For this, I developed a method called TEL Analysis. TEL stands for Toil, Effort, and Lift.

1. Toil: You ask the operators—the people doing the work—to rate the "pain" of each workflow on a scale of 1 to 3.
2. Effort: You ask the implementers (i.e., your engineers) to rate the technical effort required to automate that workflow, again on a scale of 1 to 3.
3. Lift: You ask the leaders to rate the strategic lift the company would get if you could scale that workflow without a linear increase in cost.

Now, the trick is to look for the inverse relationships.

• Workflows with a high Lift and low Effort are your low-hanging fruit for boosting enterprise value.
• Workflows with a high Toil and low Effort are your low-hanging fruit for saving money and improving team morale.

This simple, structured analysis gives you a powerful, data-driven filter to prioritize your AI initiatives, ensuring you are always focused on the projects that will create the most value in the shortest amount of time.

This three-phase process, Catalog, Classify, and Prioritize, is the blueprint for a successful AI strategy. Yes, you can do this with a series of spreadsheets. But this entire methodology is what we have engineered into our Orbit platform to turn months of manual process into a matter of a few workshops.

At the end of this process, you have a complete blueprint sorted by least effort to highest value. This might look overwhelming. The key is to think big, but start small.

Select a single top-of-the-list workflow identified by your TEL Analysis. Your goal is not to immediately build a perfect, fully autonomous L3 agent. Your goal is to run a single, focused, time-boxed sprint to create a "Minimum Viable Automation" for that one process. In most cases, you can build this automation in a day or even less.

This first sprint allows you to test your assumptions, build momentum, and deliver a tangible, measurable ROI in a short window of time. More importantly, it gives you the opportunity to create value and learn how to communicate the AI journey to the whole organization with early wins. This is how you turn your strategic blueprint into a real-world, value-creating engine.

In our next piece, we will discuss how to take this prioritized roadmap and build a sophisticated, C-level business case by quantifying the two primary forms of AI-driven value: cost savings and new opportunity creation.