← Back to Resources
    Resource · Interactive guide

    The 30–60–90 Day AI Plan

    Map AI to your real work. Three checkpoints, twelve weeks, no fluff. The workshop guide we run with founders and operators, now an interactive planner you fill in on this page.

    Last updated: 17 August 2026

    Most AI plans fail because they start with the tool

    A licence gets bought, a few people try it, and three months later nobody can say what changed. Not because the tool was wrong — because nothing was ever pointed at a specific task, on a specific date, with a specific person accountable for it.

    This plan starts with your week instead. You pick one task that genuinely costs you time, you run it for thirty days, and then you make an honest call about whether it earned a permanent spot. Twelve weeks later you either have something that is now standard practice, or you know exactly why it isn't — which is also useful.

    How to use this guide

    01

    Fill it in while it's fresh

    Complete what you can within the week of the workshop. Waiting a month means guessing at your own week instead of remembering it.

    02

    Work the horizons in order

    Each horizon builds on the one before it, so work through Day 30 before you jump to Day 90. Day 90 is a decision you earn, not a forecast.

    03

    Be specific or it won't hold

    Every prompt asks for something specific. If your answer is vague, the plan won't hold up in twelve weeks — vague plans quietly expire.

    STEP 00

    Map your real work

    Before you touch a tool, get specific about what's actually eating your time. List five recurring tasks from your week, however small. Be honest about how often you do them and how much of the work follows a pattern.

    Then circle the task with the highest time cost and the highest repetition. That's your Day 1 pilot. Not the most exciting task, not the one that would impress people — the one that costs you the most and changes the least.

    Task / activityHow oftenTime cost / weekHow repetitive?
    Weekly client status updatesWeekly3 hrsHigh
    Proposal first draftsFortnightly2 hrsMed

    An example, not a template answer. Highest time cost × highest repetition is your Day 1 pilot.

    Rate repetition as High (almost identical every time), Med (same shape, different details) or Low (varies a lot).

    The three horizons

    Each horizon has one job and five questions. Work them in order — Day 90 is a decision you earn from the first sixty days, not a forecast you write on day one.

    PHASE 01

    Days 1–30 — Foundations

    Goal. Get comfortable with how AI actually works for your role, run one pilot on a real task, and produce one result you can show someone.

    If you lead a team: pick a task from someone else's role too, not just your own. You want proof it works beyond you before you ask the team to trust it.

    What this horizon asks you

    • —The task you're piloting
    • —The tool or system you'll use
    • —What "working" looks like in 30 days — a number, if you can put one on it
    • —Who needs to know or sign off before you start
    • —Your Day 30 checkpoint date
    PHASE 02

    Days 31–60 — Build

    Goal. Turn the pilot into a habit. Either it earns a permanent spot in how you work, or you retire it and move to the next task on your list.

    If you're mapping your own role: this is where you decide what stays and what goes back to being manual. Don't keep something running just because you built it.

    What this horizon asks you

    • —Did the Day 30 pilot earn a permanent spot?
    • —If yes, who else needs to start using it
    • —The second task you'll pilot
    • —What needs to be written down so it doesn't live only in your head
    • —Your Day 60 checkpoint date
    PHASE 03

    Days 61–90 — Scale

    Goal. Decide what's now standard practice, put a number on what it actually got you back, and set next quarter's priority.

    If you lead a team: this is the point to formalise, document, and hand off. What only you can run isn't a system yet, it's a habit.

    What this horizon asks you

    • —What's now standard practice, not a pilot
    • —The time or output this actually freed up
    • —What's still manual that shouldn't be
    • —Next quarter's priority
    • —Your Day 90 review date
    Guardrails

    Human in the loop, always

    AI does the labour. People make the calls. Before you scale anything past Day 30, every one of these has to be true.

    • Someone reviews the output before it goes to a client, the team, or a decision.
    • There's a person accountable for every task you've handed to AI, not the AI.
    • You know what happens when it gets something wrong, and who catches it.
    • Nothing goes out publicly or gets acted on without a human check.
    Review rhythm

    Set all three dates now

    Set all three review dates before the session ends. A plan without dates in the calendar is just a nice idea.

    Between checkpoints, a ten-minute weekly check on the current pilot is enough. You're not auditing, you're just noticing what's working and what's quietly not.

    Fill in your plan

    The interactive version of the workshop guide. Your answers save in this browser as you type, and you can print the finished plan or copy it as text.

    Free unlock

    Unlock the interactive planner

    • The Step 00 task table, with your Day 1 pilot worked out for you
    • Every Day 30, 60 and 90 prompt as a fillable field
    • The guardrails checklist and your three review dates in one place
    • Autosave in your browser, plus print-to-PDF and copy-as-text

    Frequently asked

    Next step

    Want help running the twelve weeks?

    We train teams to use AI on their real work, then build the systems that keep it running once the plan is signed off. Book a free 30-minute consultation.

    Book a free consultation