You can see potential for AI in your organization without knowing where it should go. A useful place to begin is a process your team already understands: how you respond to an inquiry, prepare a proposal, onboard a customer, coordinate a project, or manage a renewal.
Look at the work inside that process. Find the steps where people spend time interpreting information, searching for context, or creating something another person needs. Then assess whether AI could help and whether that part of the process is ready.
This guide gives you a first pass you can use with almost any process. The result is a short list of opportunities and a clearer next step for each.
Choose a process with a recognizable beginning and end. “From a customer request to a resolved issue” is a workable boundary. “All of customer experience” is too broad for this exercise.
Ask the people involved to describe the process as it actually works. What starts it? What result should it produce? Who depends on that result?
Write down the main steps in plain language. You do not need a detailed process map. A short list is enough to begin. Include the handoffs where work moves between people or teams.
For each step, note who does the work, what information they need, and what they pass on. You are looking for a few useful opportunities within the process.
Walk through a recent example with the person who did the work. Ask where they had to read, search, compare, interpret, or write before they could move forward.
Pay particular attention to these patterns:
These are places to investigate. Repetition alone does not make a task suitable for AI. Sometimes the friction comes from unclear ownership or a missing connection between systems.
| Area | Process you could examine | A step where AI may help |
|---|---|---|
| Relationship management | Responding to a new inquiry | Organizing the person's request and relevant relationship history for review |
| Revenue management | Preparing an opportunity for the next stage | Summarizing confirmed requirements and unresolved questions |
| Retention management | Preparing for a renewal conversation | Bringing relevant feedback, service history, and open concerns into a review brief |
| Operations | Coordinating a project handoff | Drafting a status summary from approved updates and surfacing missing information |
The same approach applies elsewhere, including employee onboarding, supplier management, and internal requests. Begin with the process, then inspect the work inside it.
For each area of friction, describe what would become easier or more reliable. Be specific about the result you want.
AI may be worth exploring when the step requires interpreting language, organizing varied information, or producing a draft that someone can evaluate. A fixed rule may be sufficient when the answer is already explicit. A territory-based assignment, for example, may only need a reliable field and a workflow.
Ask four questions:
If the team disagrees about what should happen at that step, clarify the process first. Automating an unresolved decision can spread the confusion.
Use this format:
In [process], [role] spends effort [doing a task]. AI could assist with [a defined output] so that [a useful outcome]. [Role] would review the result before [the next action].
This keeps the opportunity small enough to assess. “AI could assist with preparing an internal handoff summary” describes a bounded task. “AI will run our operations” leaves too many decisions unanswered.
Write a few candidate opportunities if several steps stand out. You do not need to select one immediately.
Assess each candidate separately. One process can contain an opportunity that is ready to explore and another that needs preparation.
For each check below, record Verified, Gap, or Unknown, along with the reason. An unknown is something to investigate.
| Readiness check | What you need to know |
|---|---|
| Purpose and outcome | The team agrees on the problem, the intended output, and what useful improvement would look like |
| Process and ownership | The step has a clear owner, a recognizable trigger, and a defined next action |
| Information quality | Required information is present, consistent, and current enough for the task, with clear record identity and duplicate issues understood |
| Connections and context | Relevant information connects to the right person, organization, transaction, project, or request, and its business meaning is clear |
| Access and suitability | The intended use of the information is approved and the right people and tools can access it within appropriate limits |
| Technology fit | There is a plausible way for an appropriate AI tool to support the task and reach the required information, subject to verification |
| Human review | A named person can judge the result, handle exceptions, and approve consequential next steps |
| Evaluation and capacity | The team can define success and has time to review the work and learn whether the approach helps |
You do not need to design the solution to answer these questions. If a technical or access question remains unresolved, record it as unknown and identify who can resolve it.
This is a readiness screen for a specific opportunity. It is not a certification that a process is ready for autonomous AI.
An AI task depends on more than having fields filled in. The information must be correct enough for the work, connected to the right situation, and available to the tool performing the task.
Consider a project handoff. A status field says “Ready for the next phase,” but a recent update says a required approval is outstanding. A summary based on the status field alone could send the next person in the wrong direction.
The gap involves freshness, context, and ownership. Someone needs to reconcile the information and maintain it as the work changes.
Use the same questions in any process: What would a capable person need to know here? Where does that information live? Does it describe the current situation? Could the intended tool actually use it?
A record’s latest modified date does not prove that every fact in the record is current. Review the information that affects the task.
Compare the candidates using benefit, readiness, and review effort. Favor a useful, bounded task where the team can access the necessary information and evaluate the output.
| What you find | What it means | Your next step |
|---|---|---|
| Clear benefit, adequate information, and manageable review | A promising candidate for further validation | Explore a limited, supervised pilot |
| Clear benefit with important process or data gaps | A promising opportunity that needs preparation | Identify the gaps and assign owners |
| A fixed rule or simple process change solves the problem | AI may add unnecessary complexity | Pursue the simpler improvement |
| Unclear benefit or mistakes the team cannot manage | A weak first opportunity | Reconsider the scope or examine another step |
Avoid adding up readiness answers into a single score. A strong benefit cannot cancel out inappropriate access or an inability to check the result.
A first review can identify a promising starting point. Further validation must establish whether the chosen tool and approach work in your environment.
Capture what you learned in a simple table. Three candidates are often enough to make the next conversation useful.
| Opportunity | Expected benefit | Readiness finding and evidence | Owner | Next decision |
|---|---|---|---|---|
| The specific step and output AI could assist with | The improvement you want to see | What is verified, missing, or unknown | Who owns the next step | Explore, prepare, simplify, or defer |
If you find gaps, identify the few that matter most to the opportunity. State the question or repair needed, who owns it, and when it should be reviewed.
You now have a basis for discussing AI with your team or an implementation partner. You can explain where it may help, why the opportunity matters, and what must be true before moving forward.
You can apply this review to one process at a time. The aim is to find manageable opportunities where AI could improve the work and understand what those opportunities need.
That is how a broad interest in AI becomes an intentional operational decision.