The question
your best people
are already
asking.
Medium and large companies are primarily asking, “How do we achieve measurable ROI and tangible business value from AI investments?” amid high failure rates and massive spending.
We run a half-day workshop that surfaces where AI creates the most value in your workflows—then we can build the solution that fits what you find. The tools are a commodity. Everyone has access to the same base models. What separates organizations is the judgment to use them well — and whether the structure exists to act on what emerges. That gap is still closeable. It will not be for long.
Read the leadership argument · Half-day format · Your team · Prioritized output · Businesses of all sizes
What does your organization become when its people stop being constrained by what one person can know, synthesize, and act on alone?
AI is an accelerant: it executes instructions precisely but does not question your assumptions. Without clear direction, it amplifies noise as easily as signal—which is why thoughtful workflow choices and leadership judgment matter more than which model you pick.
Between organizations where people think with AI every day — and those where it’s still a novelty.
AI projects have high failure rates—typically 80% to 95% across general AI initiatives and even higher (up to 95%) for generative AI pilots. Only 5–20% succeed in delivering business value or scaling to production.
Main Failure Reasons
- Weak data foundations, silos, or unclear ownership.
- No shared strategy: goals, sequence, or definition of success.
- Skills, integration, or adoption lag behind the hype.
Main Success Factors
- AI‑ready data and realistic scope.
- Clear business case, metrics, and executive alignment.
- Iterative pilots tied to real workflows—not slide decks.
Most AI projects fail because they skip this step.
They never lock in where to start, what to do first, and how they will know it worked before budget, vendors, and pilots take over. Without that shared picture, every demo chases a moving target—and the organization burns time on tools instead of on decisions.
They buy the tool before understanding the problem. They hire the vendor before mapping the workflow. We fix the planning layer first: a half-day workshop with your team, the AI Opportunity Guide if you want to work through it yourself, or the free 15-minute assessment when you need a fast read on one high-impact workflow.
The tools are a commodity. A small business in Thunder Bay has access to the same AI models as the largest firm in your industry. The advantage is not access — it is knowing which workflows to point AI at, and in what order. That is what those three entry points are for.
A half-day workshop.
A prioritized list.
A starting point to act on.
We run AI opportunity identification workshops with your team. The output is specific to your workflows — not a report to file, but a prioritized list with named internal owners and a clear next step for each item.
This work requires one identified lead from your business — not a full-time role, but someone who can communicate clearly between the people doing the work and the people making decisions. That person becomes the internal engine that keeps momentum after we leave.
Prioritized AI opportunities specific to your workflows, ranked by impact and ease of implementation
Named internal owner for each opportunity — accountable and ready to move
Separation of signal from noise — what is real versus what is overhyped in your specific context
A conversation about growth — what happens when your people become more capable than the structure around them
Momentum — the kind that compounds over time rather than stalling in committee
Not ready for a half-day workshop yet? Start with a 15-minute assessment.
The AI Opportunity Workshop is a paid, team-based session: we map multiple workflows, prioritise opportunities, and leave you with named owners and next steps. It is the right move when you are ready to commit time in the room.
This free assessment is different: a short, structured conversation focused on one high-impact workflow where AI could save time in your business. You get a clear, personalised summary you can act on yourself—or use as a starting point if you want help scoping or building.
Many teams implement ideas from the report on their own. When integrations, governance, or delivery capacity are the bottleneck, that is where Optimaiz typically engages. Start a conversation when you are ready—no obligation.
The AI Opportunity Guide — the same planning sequence, on your timeline.
Five structured parts, purpose-built prompts, space to capture your answers, and a prioritisation framework so you know what to tackle first without committing to a workshop yet. Use it solo, bring it into leadership meetings, or bring a completed copy to a half-day session so the room starts aligned.
The full printable guide includes expanded prompts, write-in fields, and a “first 30 days” checklist—ideal when you want depth without scheduling a call.
Buy the guide — $19.99 USD Preview the workshop overview & guide taste
Find out exactly where AI
can save your business time
— in 15 minutes.
A conversation with a senior AI advisor. Free. No registration required. Your personalised report arrives at the end. It complements our half-day workshop—focused on one workflow, not a full team working session.
We identify the one workflow that can show results fastest — so your team sees the impact without a big project.
Start the Assessment →Scroll to learn more · or jump straight in
Video — 90 seconds · What the assessment does & what you’ll receive
This assessment is a structured AI-powered conversation that identifies where automation can create the most value in your specific business. It takes about 15 minutes, it’s free, and at the end you receive a personalised report with actionable recommendations.
Start by telling us what kind
of business you run.
The Conversation
From a free clarity talk to a full AI partner retainer.
We work with owner-operated businesses and professional services firms of all sizes. No enterprise budget required.
The window is open.
It will not stay open indefinitely.
The businesses that will look back on this period well are building something that is hard to name and easy to recognize in hindsight: a culture where human judgment and AI capability compound together, where the best people are doing their best work, and where the business has evolved to support them.
If your best people became the most capable versions of themselves—what changes?
Same judgment. Same experience. Same instincts. But now operating with access to research, synthesis, and scenario-building that no individual could sustain alone. They can see further. They can move faster. They can build a stronger case for action.
The real disruption isn’t that AI amplifies your industry. It’s that your best people are beginning to see what they can actually do. The question is whether your business has the processes, the culture, and the agility to let them do it.
The narrative runs in one direction: leaders must prepare their people for AI. The more uncomfortable truth is that leaders must prepare themselves for what their people are becoming.
A workforce that is more capable, more informed, and more autonomous does not need less leadership. It needs different leadership—faster, more transparent, more willing to act on good thinking regardless of where in the hierarchy it originates.
AI is becoming more effective, less costly, and easier to use. Workflows that don’t work today may work in six months. Plan accordingly.
The discipline is learning to pause and ask: which parts of this workflow should AI own? That habit does not develop automatically.
You don’t need an AI department. The people doing the work know which tasks AI can take over. Steering committees deliberate. People on the ground act.
Things change quickly, all the time. Any specific tool recommendation has a short shelf life. How you think about AI in your context has a longer one.
The hardest emerging management question: what do you do with people who are now capable of more than the role you hired them for? For your best people, this is already here.
Pattern recognition.
Not a roadmap.
No one has an honest roadmap to sell you. We arrive with pattern recognition from working across businesses at different stages of this transition — and a clear process for identifying where AI creates the most immediate value in your specific workflows.
What does your organization become when its people stop being constrained by what one person can know, synthesize, and act on alone?
The Question Your Best
People Are Already Asking
The tools are a commodity; access is not the moat. The competition has already shifted to judgment, workflow choices, and leadership clarity. You already believe AI matters—that argument is over. What most leaders still lack is the next sentence after ‘okay, so now what?’
The sections above are about planning and starting well. What follows is the organisational and leadership read: what changes when your best people become more capable than the structure around them.
What is not a commodity
is the judgment to use it.
What is not a commodity is the judgment to know what to ask of these tools, the experience to recognize which tasks in a workflow can be handed off to an AI, and the relationships that turn good analysis into real decisions inside complex organizations. Those things are human. They compound over time. They cannot be downloaded.
AI in workflows should come from the people doing the work — those who have a real need to get more done, better. They know which tasks AI can take over. They may not know how. That is where outside perspective is genuinely useful: not to replace their judgment, but to bring pattern recognition across organizations. What we bring is pattern recognition — we have seen what works and what doesn't across multiple organizations in real time.
The gap opening right now is not between organizations with AI access and those without. It is between organizations where people are genuinely thinking with AI every day, and those where it remains a novelty used occasionally by the curious.
That gap compounds. The longer an organization waits, the harder the catch-up becomes. It is still closeable now.
Everyone has access to the tools. Not everyone is using them. Almost no one is using them to ask the questions that actually matter.
If your best people became the most capable versions of themselves — what changes?
Here is the question worth sitting with. If your people — not all of them, but your best ones — began working with AI seriously enough that they became, functionally, the most informed and analytically capable versions of themselves they have ever been, what changes?
Not different people. Same judgment, same experience, same instincts. But now operating with a depth of research, synthesis, and scenario-building that no individual could previously access alone. They can see further. They can move faster. They can build a stronger case for action than any single person in that role has ever been able to build before.
So what do they do with that on Monday morning?
The honest answer: it depends entirely on whether your organization has thought about this question before it arrives. Because it is arriving.
Leaders must prepare themselves for what their people are becoming.
The real disruption isn't only that AI amplifies your industry. It's that your best people are beginning to see what they can actually do — more, better, faster than before. The question is whether your organization has the processes, the culture, and the leadership agility to let them do it. The organizations that don't make room for this capability will train their competitors.
This cuts in both directions, and that is the part that rarely gets said plainly. Your best people still need leadership. They need vision and an environment within which they flourish amidst constant change. They need a CEO. But they need a better CEO than they needed before — not because the previous one was deficient, but because the people being led have changed. The organization that figures this out first will attract and keep exactly the people it needs.
The narrative almost always runs in one direction: leaders must prepare their people for AI. That is true and insufficient. The more uncomfortable truth is that leaders must also prepare themselves for what their people are becoming. A workforce that is more capable, more informed, and more autonomous does not need less leadership. It needs different leadership — faster, more transparent, more willing to act on good thinking regardless of where in the hierarchy it originates.
Decisions made closer to the information rather than escalated up the org chart. Analysis that comes from a junior hire acted on because it is good, not set aside because of who produced it. A tolerance for moving before the picture is complete, because waiting for certainty in a fast-moving environment is itself a decision with consequences.
The question is not whether your people are ready for AI. Some of them already are. The question is whether the organization is ready for them.
Capable Isn't Enough.
Your AI Needs to Know What You Want.
The race to deploy AI is producing a predictable problem: organizations handing authority to systems that have never been told what actually matters. A machine that executes efficiently against the wrong objective is not an asset. It is a liability that scales.
The thinking in this piece belongs to Nate B Jones. His work on intent engineering and organizational AI architecture is where these ideas originate. We've interpreted and reframed his framework here for a leadership audience — but the intellectual foundation is his.
A competent agent doing the wrong thing at scale
Consider what happens when an AI agent is deployed into customer service with instructions to resolve issues quickly. It does exactly that. Handle rates climb. Average resolution time drops. Every metric the business is watching moves in the right direction.
What the metrics don't capture: the agent is closing cases in ways that erode the relationship. It is technically correct and strategically wrong. Nobody told it the difference, because "retain loyal customers" was never translated into something a machine could act on. The agent had capability without direction — which turns out to be its own kind of problem.
This is not a failure of the technology. It is a failure of the mandate. And it is more common than most leadership teams realize.
Organizations have always operationalized strategy imperfectly. A new hire interprets their role through instinct, peer behaviour, and whatever their manager emphasizes this quarter. Imprecision is the normal condition of any large organization. Why does AI require more explicit direction?
Because human misalignment corrects slowly. AI misalignment scales instantly. A person making the wrong call affects a handful of situations before a manager catches it. An AI making the same wrong call touches thousands of interactions before the pattern surfaces — by which point the damage is already baked into your churn numbers. The organizational immune system that catches human drift simply does not operate at machine speed.
Human goals are not machine instructions
Strategic goals live comfortably in human language. They do not survive contact with an AI system. Before any agent is given authority, four things must exist in explicit, machine-readable form: the measurable signals that define success, the data sources where those signals live, an enumerated list of permitted actions, and the hard boundaries it may never cross regardless of circumstances.
Without these, the agent will optimize for whatever is measurable and proximate. That is rarely what the organization actually cares about.
The infrastructure nobody is building yet
Experienced leaders carry organizational judgment the way they carry their own instincts — accumulated, contextual, rarely articulated. Which clients get exceptions. Where the real risk tolerance sits, as opposed to the stated one. When a policy should bend. None of this is written down because it has never needed to be. Humans read context. Machines cannot.
The work of this period is making that implicit knowledge explicit — translating it into parameters an AI can actually use. Most organizations have not started. The constraint on AI performance right now is not the capability of the models. It is the absence of this layer.
There is a harder problem underneath this one. Encoding organizational judgment assumes the organization has settled, consistent judgment to encode. Many do not. Priorities shift. Leaders disagree. What gets said in strategy sessions and what gets decided under pressure are often different things.
Building an intent architecture will surface those contradictions — and that is precisely why it is worth doing. The process forces resolution of ambiguities that have been managed quietly for years. An AI left to navigate that ambiguity on its own will resolve it too — just not in any way you deliberately chose.
The model is not the advantage
Every organization has access to roughly the same AI models. That access will only become more equal over time. The technology itself is not a durable advantage — it is infrastructure, like cloud computing or broadband before it.
What cannot be bought off the shelf is the layer that sits beneath the model: the encoded expression of what your organization is actually trying to do, what it refuses to do, and how it makes trade-offs when those things are in tension. That layer comes from the inside. It takes time to build. And unlike the model, it compounds — each deployment teaching the organization more about where its judgment needs to be sharper.
A practical test: Pick your next planned AI initiative. Can your team answer all four questions in the framework above — specifically, in writing, with named data sources? If not, the initiative is not ready to deploy. Proceeding without those answers is not moving fast. It is moving without a destination.
This is a leadership problem, not a technical one. It requires the senior team to do something genuinely difficult: reach explicit agreement on what the organization is optimizing for, what it will not compromise, and how those answers get maintained as the business evolves. No vendor solves that. No model does it for you. It is the work that makes everything else work.
When the workshop points to a concrete workflow, we can build it. One current focus: AI-assisted proposal writing.
This is not a separate product pitch at the top of the funnel—it is an illustration. When proposals surface as a top opportunity for your firm, a build might look like the multi-stage proposal intelligence workflow below. Expand only if you want the technical and commercial detail.
AI opportunity identification.
Overview and preparation — a taste of the AI Opportunity Guide.
This page gives you an overview of the workshop and preparation templates for business-unit leads. It’s the same approach we use in the AI Opportunity Guide—a step-by-step booklet you can work through on your own. Complete the templates before a session so we can focus the discussion on where AI can improve your workflows and identify quick wins; or use them on your own to scope one process. Any industry—professional services, operations, finance, or elsewhere—can use this approach.
If you’d like to go deeper on your own, the AI Opportunity Guide booklet is more comprehensive: the same five steps, with many more prompts, dedicated write-in space for your answers, and a First 30 days checklist.
You can save this page as HTML or use your browser’s “Print to PDF” to keep a copy.
What this workshop is and who it’s for.
The AI opportunity identification workshop is a half-day session with your team and key people from your main business areas. The goal is to surface task-level opportunities where AI can improve workflows — not to sell tools, but to identify where better judgment, speed, or consistency would create real value.
We focus on quick wins: changes that can show results in weeks or a few months, so your business builds momentum instead of waiting for a large-scale rollout. A quick win usually has: bounded scope, data you already have, one clear owner, and a result you can measure in weeks.
Participants are asked to detail their workflows in discrete steps in advance using the preparation templates on this page. That prep becomes the basis for the discussion—where the pain points are, where rework happens, and where a “back of the napkin” improvement would be worth pursuing.
Asking you to break down your workflows is difficult—and that’s the point.
Most teams describe their work at a high level: “We do intake, then we process, then we deliver.” That’s not enough. The real opportunities show up when you get to the nitty-gritty: the specific trigger that starts the process, who does what at each step, what data or documents move between steps, where handoffs happen, and where time or errors pile up.
Breaking workflows into discrete steps is hard because it forces honesty. You have to describe the process as it really runs—not the ideal version. That’s where bottlenecks, hidden waste, and AI opportunities become visible. The templates on this page are designed to push you to that level of detail. The more you put in, the more we can identify concrete, actionable opportunities in the room.
Use the templates thoroughly. Process name, owner, start trigger, end output, step-by-step breakdown, key people, data and docs used, volume, time per run, and top pain points. This level of detail is what separates a vague wish list from a prioritized list you can act on.
How we structure the exploration.
We follow a clear sequence so the workshop stays focused and produces comparable, prioritizable output.
Everyone comes having completed (or started) the preparation templates. In the room we work through what you’ve outlined, stress-test which opportunities are realistic quick wins, and leave with a prioritized list and named owners.
One person from your business should be identified in advance as the internal lead — someone who can coordinate between the people doing the work and the people making decisions. That person keeps momentum after we leave.
How organizations are using AI in workflows.
These patterns show up across industries. The shift is underway at the workflow level—not the strategy-deck level.
Document intake & classification
From unstructured inputs to structured data
Organizations are using AI to receive PDFs, scans, emails, and forms; classify document type; and extract key data into systems. What used to be manual opening, reading, and re-keying is now assisted by document understanding. The gain is time and consistency—people focus on exceptions and judgment while routine intake is handled.
Routing & triage
Getting work to the right place without manual handoffs
Where work is assigned or routed by people, delays and wrong routing are common. AI can classify incoming work, suggest or assign the right team, and reduce handoffs. The result is fewer queues and faster throughput. This pattern applies whether the “work” is a claim, a request, a proposal, or an order.
Knowledge retrieval & synthesis
Finding and summarising what already exists
Teams spend a lot of time searching for past examples, precedents, or answers that already exist in documents or systems. AI-assisted search and summarisation can surface the right information at the right time. The value is in freeing skilled people from retrieval work so they can focus on analysis and decisions.
Prompts to help you think about where AI can help.
Here are a few prompts we use across four phases: map the process, find the pain, imagine the AI, then test feasibility. Reflecting on them before the workshop makes the discussion more productive. The AI Opportunity Guide booklet includes a much fuller set of prompts for each phase—with space to write in your answers—plus the four essential questions you answer for each opportunity (what success looks like, what understanding the AI needs, who reviews, and what the AI must never do). If you find these useful, the full guide will take you further.
- Walk us through your core process end-to-end. What is the first trigger and what is the final output?
- What data or documents does this process consume? Where do they come from and what format are they in?
- If you had to name the ONE step that is most painful, slow, or error-prone—what is it?
- Where do mistakes most commonly happen—and what’s the cost when they do?
- If you could eliminate one manual step completely—which would it be and why?
- Where do your best people spend the most time—and could AI free them to do more of that?
- Do we have historical data for this process—and is it clean enough for an AI to learn from?
- What would success look like in 6 months—specifically and measurably?
What you leave with.
The workshop is designed to produce concrete deliverables, not general advice.
Prioritized AI opportunities specific to your workflows, ranked by impact and ease of implementation
Named internal owner for each opportunity—accountable and ready to move
Quick-win candidates that can show results in weeks or a few months
A clear next step for each item so the session leads to action
Want the full experience? This overview and these templates give you a helpful head start. The AI Opportunity Guide booklet is more comprehensive: the same five parts with a fuller set of prompts at every step, dedicated write-in space for your process, the four essential questions for each opportunity (with room to answer), governance and who does what, and a First 30 days checklist to turn the guide into action. You can work through it on screen or print it and write by hand.
Workflow breakdown.
A short preview of the workflow form. The AI Opportunity Guide booklet has the full version with more space and the same fields so you can pick up where you left off. Describe the real process, not the ideal one.
Same structure as Part 1 and Part 2 in the booklet: scope the process, then break it into steps.
| # | What happens | Who does it | Inputs / outputs | Rough time or frequency | Where could AI help? (optional) |
|---|---|---|---|---|---|
| 1 | |||||
| 2 | |||||
| 3 | |||||
| 4 |
In the full AI Opportunity Guide you get more write-in space, “If you’re stuck” prompts, and the same labels so this preview maps directly to the booklet.
Pain, value, and AI opportunity.
A preview of the pain-and-opportunity form. The AI Opportunity Guide booklet has the full version with the same labels and more write-in space (Parts 3, 4, and 5), plus the four essential questions and prioritisation grid.
Same structure as in the booklet so you can continue there without relearning the form.
| Opportunity | Impact 1–5 |
Feasibility 1–5 |
Data ready? 1–5 |
Speed to value 1–5 |
Total |
|---|---|---|---|---|---|
| Your opportunity | |||||
| Your opportunity |
In the booklet you also get: four essential questions (what success looks like, what understanding the AI needs, who reviews, what the AI must never do), plus governance and First 30 days.
Bring this to the workshop or use it as a quick scan. For the full experience, use the AI Opportunity Guide booklet.
AI Opportunity Guide
Start your journey on your own.
A step-by-step booklet to help you scope a process, break it into real steps, find where it hurts, and identify where AI could create quick wins. Fill it in on screen or print it and write by hand. No prior AI experience required.
Enter your name and email to access the booklet.
Those credentials don’t match our records. To access the booklet, please use the button below to purchase.
To buy this booklet click hereWork at your own pace. There are no wrong answers.
Where to begin. The strongest place to start is a task your team already does often—and that takes longer than it should. You don’t need a full strategy before you start. You need a concrete task, a tool (or a willingness to try one), and someone who will run the experiment. It works best when completed by someone who does or oversees the process day to day—they know where it really hurts.
This guide walks you through five parts: Scope your domain, Break down your journey, Find pain and value, Identify AI opportunities, and Next steps. Each part has space to write your answers and, if you get stuck, short prompts to nudge your thinking. For each opportunity you identify, you’ll also work through four check questions later in the booklet. If you can’t answer all four clearly, that opportunity isn’t ready to move forward yet.
You can complete it for one process or workflow. If you have several, pick the one that matters most or that you know best. Honesty beats perfection: describe how the process really runs, not how it should.
At the end you’ll find a First 30 days checklist to turn your notes into action. You can return to this booklet whenever you add a new workflow or want to compare opportunities.
Scope your domain.
Choose one process or workflow to focus on. Naming it and defining its start and end keeps everything else bounded. Who is responsible for it? What kicks it off, and what is the final output?
Name the process
Give it a clear name (e.g. “Customer onboarding,” “Invoice processing,” “Proposal development from RFP to submission”).
Ask yourself: What is the very first thing that has to happen before anyone does any work? Is it an email, a form, a date, a request from another team?
Break down your journey.
This is the hardest part—and the most important. List every major step in order. For each step, note what happens, who does it, what goes in and out, and roughly how long it takes. Describe the real process, not the ideal one. Include the messy, manual, or duplicated steps. Someone outside your team should be able to follow the sequence.
Tip: If a step is vague (“we process it”), break it into smaller steps. “We open the email, we read the attachment, we type the key fields into the system, we send a confirmation” is better than one big “we process it.”
| # | What happens | Who does it | Inputs / outputs | Rough time or frequency | Where could AI help? (optional) |
|---|---|---|---|---|---|
| 1 | |||||
| 2 | |||||
| 3 | |||||
| 4 | |||||
| 5 | |||||
| 6 | |||||
| 7 | |||||
| 8+ |
Walk through one recent example from start to finish. What was the trigger? What did the first person do? Then the next? Where did things wait or get handed off? Write down what actually happened, step by step.
- Where does work sit in someone’s inbox or queue?
- Where do people copy information from one system to another?
- Where does one person’s knowledge become a bottleneck?
- How is work assigned or routed to the right person or team today?
- Which steps are rules-based and predictable? Which require judgment?
Find pain and value.
Where does it hurt? Where do mistakes happen, delays build, or people get frustrated? Be specific. Then ask: if we improved this, what would it be worth? You don’t need exact numbers—a back-of-the-napkin estimate is enough (e.g. “20% less time,” “fewer errors in step 3,” “2 weeks faster to delivery”).
Ask: When something goes wrong, where does it usually go wrong? When customers or colleagues complain about this process, what do they complain about? Where do backlogs or queues build up?
- When does your process involve reading a document and extracting specific information? (That’s often prime territory for AI.)
- What decisions does your team make repeatedly that follow a similar pattern each time?
Identify the AI opportunity.
The steps that hurt most or feel most repetitive (from Part 3) are often the best candidates. Which of those could be assisted by AI? Think about the type of help: reading and extracting information from documents, summarising text, routing work to the right place, answering repeat questions, or spotting patterns. You don’t need to know the technology—just the kind of task.
Common patterns: Document intake and classification · Extracting data from forms or PDFs · Summarising long documents · Routing or triaging work · Answering repeated questions from a knowledge base · Spotting anomalies or risks. Which of these (or something else) might fit your process?
Look back at your step list and the “Where could AI help?” column if you filled it in.
Look back at your step list. For each step, ask: Could a tool read documents here? Could it suggest an answer or a next action? Could it do the first draft so a person only reviews? Could it route work to the right person automatically?
- If you could eliminate one manual step completely—which would it be and why?
- Which steps could theoretically run without a human, and which couldn’t?
- Where do your best people spend the most time—and could AI free them to do more of that?
- What data does your process generate that you’ve never had time to analyse?
What makes a quick win: Bounded scope, data you already have, one clear owner, and a result you can measure in weeks.
Next steps and prioritisation.
Before you move an opportunity forward, run it through the four questions below. If you can’t answer each one in a specific way, the opportunity isn’t ready yet—refine it and try again. If you have more than one opportunity, run these four questions for each—use extra paper or copy this section so each opportunity has its own answers.
If you can’t answer these clearly, or you don’t have data or an owner, pause and refine the opportunity (or pick another) before piloting.
- No data (or data not in a usable form)
- No clear owner for the pilot
- Can’t define what success looks like in measurable terms
Essential questions
1. What does success look like?
Define the measurable evidence of success—not the high-level goal, but the signals you could actually track. If you can’t name them, the opportunity isn’t defined well enough yet. Example: instead of “improve customer response quality,” use: fewer follow-up contacts, a specific satisfaction score, or time to first draft down from 2 hours to 15 minutes.
2. What understanding does the AI need?
What documents, data, policies, or examples does the AI need to produce something useful? Context is what turns generic output into something you can use. Without it, results are weak; with the right context, they become actionable. Think: templates, sample outputs, audience and tone, relevant precedents, product or service documentation.
3. What is the AI allowed to do—and who reviews?
Spell out what the AI can do and who must review the output before it is used or published. This should be an explicit list, not a vague sense of what seems okay. Example: the AI drafts; a named staff member reviews and approves before anything is sent. Or: the AI summarises internal documents only—no external use.
4. Do not allow the AI to…
Define the hard limits: sensitive data, legally protected information, decisions that must stay with a human. These are non-negotiable boundaries. Examples: never put customer personal data into unapproved tools; never publish AI output without human review; never use AI to make final decisions on matters with legal or regulatory impact.
Turn your thinking into action. What is the smallest experiment to test this idea? Who would own it? What would success look like in 6 months—in one sentence, and measurable?
| Opportunity | Impact 1–5 |
Feasibility 1–5 |
Data ready? 1–5 |
Speed to value 1–5 |
Total |
|---|---|---|---|---|---|
Scores 20+ = consider starting a pilot soon. 15–19 = plan for the next quarter. Below 15 = revisit when data or capacity is ready.
How AI work actually happens.
Using AI well is a loop, not a one-off. Useful output comes from iteration—refining your request and context—not from a single perfect prompt.
Plan your result
Clarify the outcome you want. Write down the steps of your current process and what a good output looks like. The clearer your intent, the better the result.
Gather materials
Gather the right documents, policies, examples, and instructions. The context you give is what separates useful output from generic filler. Ask: what would I hand to a skilled new hire to get this done?
Review outputs
Review the output for accuracy, tone, and completeness. Always verify facts and specific claims. AI can be confidently wrong—human review is the safeguard, not an afterthought.
Iterate / Improve
Adjust your prompt, add context, or give an example of what you want. Each round teaches you how to direct the tool better. Keep the prompts and context that work.
Governance.
Before you scale any AI use, get two things clear.
- Define what the AI can and cannot do. Be explicit about scope, boundaries, and where human judgment is required.
- Define who is involved and who signs off. Be clear on who must review, approve, or be consulted—and who is accountable for the output.
Who does what.
AI adoption needs three distinct functions. They don’t have to be three different people—in smaller teams one person may cover more than one. But the functions need to exist. Identify who owns each before you finish.
Leaders: set direction and gain executive-level approvals. End Users: have the detail necessary to properly determine what an improvement actually is. Champions: are highly motivated AI champions who want to learn all they can, evangelise AI, and constantly look for ways to improve the effort.
The essential questions, the plan–gather–review–iterate loop, and these three roles are the structure. Tools, timelines, and specific use cases fill in once you know which opportunities are worth pursuing and who is responsible for each.
Turn this booklet into action.
Use this checklist to take your first steps. Tick or write the date when you complete each. You don’t have to do them in order—pick what makes sense for you.
You’ve started. This booklet is a guide, not a one-time test. Revisit it when you add another workflow, when you run your first pilot, or when you want to compare and prioritize more opportunities. For a facilitated half-day workshop with your team and a prioritized list with named owners, get in touch with Optimaiz.
Contact Optimaiz.ca · We are here to help.