Knowledge that does not become practice fades, and the most common way a program like this one fails a learner is not that the lessons were unclear but that the learner finished them, felt informed, and changed nothing about how they actually work. This final lesson of the first level exists to prevent exactly that, by converting everything you have learned, what AI is, where it helps, where it endangers, the cardinal rule, the verification disciplines, the governance landscape, into a concrete 90-day plan that moves you from awareness to a real, if early, change in practice. The plan is deliberately modest, because the goal of these first 90 days is not to transform your pharmacy or master every technique; it is to internalize the core dispositions, run one real AI-assisted task with proper verification, and end with a clear, honest assessment of where AI fits in your work and what you would build next. Awareness is the achievement of this level; the 90-day on-ramp is how you make that awareness stick by acting on it, in small, safe, real ways, before the knowledge has a chance to fade into the comfortable feeling of having once read about AI.

Days 1 to 30: See and Question

The first month is about perception, training yourself to actually see the AI already present in your work and to apply the skeptical questions this level taught, without yet changing any high-stakes practice. The goal is to make the invisible visible, because AI is often embedded in tools you already use without being labeled as such, and you cannot bring discipline to what you do not notice. Spend this month doing three things. First, inventory the AI in your daily work: notice where your tools are extracting, generating, or surfacing signals, the dispensing alerts, the prior-authorization assistant, the counseling-content feature, the summarization tool, and name which of the three jobs each is doing. Second, practice the verification questions without yet depending on them: when an AI output crosses your path, ask, even just to yourself, what kind of output is this, what would I check to verify it, and am I prepared to own a decision based on it. Third, notice your own automation bias: catch yourself in the moments you are tempted to accept an AI output because it is convenient, and simply observe that tendency, because awareness of it is the first step to managing it.

A small but powerful exercise anchors this month: pick a single AI output you encounter, anything, an alert, a draft, a summary, and trace it all the way through the framework out loud or on paper. Which of the three jobs is this? What is the load-bearing fact inside it that an error would corrupt? Where would I look to verify that fact? Am I, right now, in the habit of verifying it, or have I been accepting it? Doing this even once with full attention is more valuable than reading about it ten times, because it moves the framework from something you understand to something you have actually run, and running it once on a real output is what reveals whether your current practice is as careful as you assumed. Most pharmacists who do this exercise honestly discover at least one AI output they had been accepting without the verification it deserved, which is precisely the kind of realization the first month is designed to produce.

Nothing in this first month requires you to change a high-stakes decision or adopt a new tool; it is purely about seeing clearly and building the questioning reflex on low-stakes ground. This matters because the disciplines of the later months rest on perception: a pharmacist who cannot see the AI in their work, or who does not notice their own tendency to defer to it, cannot apply verification to it. By the end of 30 days, you should be able to look at your workflow and accurately describe where AI is, what each instance is doing, and where the verification would need to go, which is the perceptual foundation everything else builds on. Many pharmacists are surprised, during this month, by how much AI is already in their work that they had stopped noticing, the alerts that have become background noise, the autofilled fields, the suggested text, and that surprise is itself the point: the AI you have stopped seeing is exactly the AI you have stopped verifying, so making it visible again is the precondition for bringing any discipline to it at all.

The first 30 days are about perception: make the AI in your work visible, build the questioning reflex on low-stakes ground, and notice your own automation bias before trying to manage it.

Days 31 to 60: Verify One Real Thing

The second month moves from perception to practice, and the practice is deliberately narrow: pick one real, appropriate AI-assisted task and do it with full, proper verification, repeatedly, until the verification is a habit rather than an effort. The task should be one that genuinely fits the four-part shape from the helping lesson, high volume, low irreducible judgment in the AI's part, anchored to a verifiable source, with you as the human checkpoint, and one that is sanctioned and safe to practice on. For many readers a prior-authorization assembly is ideal, because it is the goldmine and because its verification, tracing the clinical facts to the chart and the criteria to the payer rules, is concrete and learnable. But the specific task matters less than the discipline: whatever you choose, the point is to run the full verification every time, deliberately, until it becomes the natural way you do that task rather than an extra step you have to remember.

A practical note on choosing the task: pick one you do often enough that you will get real repetitions within the month, because the habit forms through repetition, not through a single careful performance. A task you do once a week gives you only four chances in the month, which is rarely enough for verification to become automatic; a task you do several times a day gives you dozens, which is plenty. This is another reason prior-authorization assembly is a strong choice for many readers: it recurs frequently, so the verification gets practiced enough times to become genuinely habitual rather than remaining a deliberate effort you have to consciously summon. If your highest-value AI use is something you do rarely, it is fine to practice the verification habit on a more frequent lower-stakes task first, and carry the habit over to the rare high-stakes one once it is established, because the discipline transfers even though the specific facts differ.

This is where the abstract becomes real, and it is the most important month of the three, because a verification discipline you have actually practiced on a real task is fundamentally different from one you have only read about. You will discover things the reading could not teach you: how the AI's output actually fails on your real cases, how long the verification genuinely takes once it is a habit (less than you feared), what the AI is genuinely good at and what it consistently gets wrong in your specific context. You will also, if you are honest, catch the AI in at least one error you would have missed without the verification, and that single caught error is the moment the discipline stops being an abstraction and becomes something you believe in, because you saw what it prevented. Keep a simple record of this month's work, not elaborate documentation, just enough to remember what you did and what you found, because that record is both the seed of the documentation habit and the raw material for the month-three assessment. A useful way to structure that record is to note, for each instance, what the AI produced, what you verified, and whether you found a problem, because that triple, output, verification, outcome, is the skeleton of the verification record the governance lessons described, and practicing it now on a small scale builds the habit you will later formalize. You are not trying to produce audit-grade documentation in month two; you are building the muscle of capturing your verification as you do it, so that when a later level or an accreditation asks for real documentation, the habit of recording is already there and only needs to be made more rigorous, not invented from scratch.

Days 61 to 90: Assess and Plan

The third month steps back from doing to thinking, consolidating what the first two months taught into a clear, honest assessment and a plan for what comes next. By now you have seen the AI in your work, practiced verification on a real task, and learned concretely how the tool behaves in your hands. The third month is about turning that experience into judgment: where does AI genuinely help in your specific work, where is it not worth the trouble or too risky, what verification does each use actually require, and what would you want to build or change if you took it further. This is the L1 capstone in spirit, the "AI Opportunity and Risk Memo" the level points toward, a clear-eyed assessment of one real AI use case in your work, covering where it fits, what verification it needs, what governance or documentation applies, and a genuine go or no-go recommendation grounded in what you actually learned rather than in hype or fear.

The value of this third month is that it converts experience into a transferable judgment you can act on and articulate, which is what separates a pharmacist who has done some AI tasks from one who understands AI in their work. The assessment also sets up the next level, because the natural conclusion of an honest 90 days is usually a clear sense of one or two AI uses worth doing properly, with the verification and eventually the workflow and governance that the later levels build. You end the first level not just informed but oriented: you know what AI is, you have verified real output, you have a grounded view of where it fits in your work, and you have a concrete sense of what you would build next, which is exactly the position from which the hands-on work of Level 2 makes sense. The 90-day on-ramp, done honestly, is the bridge from knowing about AI to using it well, and crossing that bridge is the real achievement of this first level.

What the Opportunity and Risk Memo Actually Contains

Because the month-three memo is the capstone of this level, it is worth being concrete about what a good one contains, so the assessment has a clear target rather than a vague aspiration. A strong AI Opportunity and Risk Memo for a single real use case answers a handful of specific questions plainly. It names the use case and which of the three jobs the AI is doing in it. It states where the use genuinely helps, in concrete terms of time saved or burden relieved, grounded in what you observed during month two rather than what a vendor promised. It identifies the specific failure modes that apply, the wrong number, the fabricated fact, the dropped warning, the automation bias, whichever are real for this use, and states the verification that catches each. It addresses the governance and protection questions the level raised: is patient information handled safely, is the tool sanctioned and properly bound, what documentation would the use require. And it ends with a genuine recommendation: go, with the verification and conditions named, or no-go, with the reason, or a qualified "go for this narrow part, not that broader one."

The discipline of writing this memo is itself valuable, separate from the document it produces, because it forces the diffuse experience of ninety days into a structured judgment that you can defend, act on, and hand to someone else. A pharmacist who can produce that memo has done something a pharmacist who merely used some AI tools has not: they have converted experience into a clear, communicable assessment of where AI belongs in their work and on what terms, which is the exact capability the later levels build on and the exact thing an employer, a manager, or an accreditation effort will value. The memo is modest in scope, one use case, honestly assessed, but it is the genuine article: a real piece of the judgment that defines a competent AI-using professional, produced from real experience rather than received from a lecture.

How to Make the On-Ramp Actually Stick

A plan is only as good as its execution, and a few practical principles dramatically raise the odds that this on-ramp produces real change rather than good intentions. The first is to start small and real rather than large and aspirational: one task verified properly beats a grand plan to transform your AI use that never survives contact with a busy week. The second is to make the verification non-negotiable on your chosen task: the entire value of month two comes from never skipping the verification, because a discipline practiced inconsistently never becomes a habit, and the busy day when you are tempted to skip it is precisely the day the practice either takes root or does not. The third is to be honest in the month-three assessment, including about where AI did not help or where the risk was not worth it, because a real assessment that says no to a use case is more valuable than a hopeful one that says yes to everything, and the credibility of your judgment depends on its honesty.

The final principle is to remember why you are doing this, which is not compliance or career-protection in the abstract but the concrete goal this whole program serves: getting medications to patients faster without ever compromising the verification that keeps them safe. Every step of the on-ramp, the perception, the practiced verification, the honest assessment, is in service of becoming the kind of pharmacy professional who can use these powerful tools to relieve the administrative burden and serve patients better, while holding the discipline that ensures the speed never costs a patient their safety. That is the pharmacist this program is built to produce, and the 90-day on-ramp is the first real stretch of the path from awareness to that competence. You finish this level knowing what AI is and what it demands; you begin the next by building, hands-on, the workflows that turn this awareness into a practice. The on-ramp is how you make sure the knowledge of this level becomes the foundation of that practice rather than a thing you once learned and let fade, and starting it, in your real work, in small and safe and honest ways, is the single best thing you can do with everything this level has taught you. So as you close this first level, do not file it away as completed reading. Pick the day, this week, when you will start the 30-day perception phase, choose the one task you will practice verification on in month two, and put the month-three memo on your calendar as a real commitment. The professionals who get value from a program like this are not the ones who read the most carefully but the ones who act on what they read, and the gap between those two groups is almost entirely the willingness to begin. You now know enough to begin well. The only thing left is to do it.

Key Takeaways

• Knowledge that does not become practice fades; the 90-day on-ramp converts the awareness of this level into a real, if early, change in how you work, so the learning sticks instead of becoming the comfortable feeling of having once read about AI.
• Days 1 to 30 are about perception: inventory the AI already in your work and name which of the three jobs each instance does, practice the verification questions on low-stakes ground, and notice your own automation bias.
• Days 31 to 60 are about practice: pick one real, appropriate, sanctioned AI-assisted task and do it with full verification every time until the verification is a habit; this is the most important month, because practiced discipline is fundamentally different from read-about discipline.
• The caught error you would otherwise have missed is the moment verification stops being an abstraction and becomes something you believe in; keep a simple record, the seed of the documentation habit.
• Days 61 to 90 are about assessment: turn your experience into an honest "AI Opportunity and Risk Memo" for one real use case, covering fit, verification, governance, and a genuine go or no-go grounded in what you learned.
• Make it stick by starting small and real, making the verification non-negotiable on your chosen task, and being honest in the assessment, including about where AI did not help.
• The on-ramp serves the program's goal, getting medications to patients faster without ever compromising the verification that keeps them safe, and it is the bridge from knowing about AI to using it well, the real achievement of this first level.