Courses / AI Integration For Companies / Week 2

Week 2 of 12 · Beginner

Finding Use Cases

You learn to map a business process, spot the steps where AI can help, and rank ideas on value and feasibility. You leave with a ranked list that separates quick wins from larger bets.

3 hours16 slides6 min read

Open slides Read study notes

From 'AI Everywhere' To One Use Case Map The Process First Where AI Fits Inside A Process Scoring Value Scoring Feasibility The Value And Feasibility Matrix Examples And The One-Page Brief

You Will Be Able To

Map a process from trigger to result, with owners, time and error points.
Identify which steps suit AI patterns such as drafting, classifying, extracting and forecasting.
Score a use case for value and for feasibility with a simple, repeatable method.
Place ideas on a value and feasibility matrix and separate quick wins from bets.
Write a one-page use case brief that a manager can approve or reject.

From 'AI Everywhere' To One Use Case

In week 1 you saw that AI helps most when it is attached to a named task with a measurable result. Many companies skip that step. They launch a general chat tool, hope people find uses, and later wonder why nothing shows in the numbers. A use case is a specific task, in a specific team, with a specific goal. 'Use AI in customer service' is a theme. 'Draft first replies to billing emails so agents answer within four hours' is a use case.

This week gives you a method in three moves: map the work, score the ideas, and choose. The method is deliberately simple so that a finance team, an HR team and an operations team can all use the same sheet and compare results.

Map The Process First

Process mapping means writing down how work actually moves, step by step. Start with the trigger, which is the event that begins the work, such as a supplier invoice arriving. End with the result, such as the invoice paid and recorded. In between, list each step and note who does it, which system they use, how long it takes, and what goes wrong.

Talk to the people who do the work, not only their managers. The real process often differs from the written procedure, and the differences are where time hides. Look for handoffs, where work waits for another person, and for rework, where work returns because something was missing or wrong. Four signals point to promising steps:

Keep the map light. One page and an hour with the people who do the work is enough for a first pass. Perfect diagrams are not the goal.

Repetition: the same kind of work happens many times a week.
Text or documents: the step involves reading, writing or searching.
Waiting: work sits in a queue because a person must read or classify it first.
Rules of thumb: experienced staff decide quickly, but newcomers struggle.

From Idea To Ranked List

Map the process
Label AI patterns
Score value
Score feasibility
Place on matrix

Each step produces a page that others can review and challenge.

Where AI Fits Inside A Process

Most business uses of AI repeat a small set of patterns. Drafting produces a first version of an email, report or code. Summarizing shortens long material. Classifying sorts items into categories, such as routing a ticket. Extracting pulls fields from documents, such as a date and an amount from an invoice. Searching answers questions from a body of company documents. Forecasting predicts a number, such as next month's demand. Flagging marks unusual items for a person to review.

Notice what is missing from the list: making final decisions about people, money or safety without review. In early projects, AI prepares, suggests and sorts, and a person decides. That keeps risk low and builds trust. When you read your process map, label each step with one of these patterns. A step that fits none of them is probably not a first candidate.

Seven AI Patterns You Can Look For

Pattern	What it does	Example
Draft	Writes a first version	Reply to a billing email
Summarize	Shortens long material	Ticket history before a call
Classify	Sorts into categories	Route a request to a team
Extract	Pulls fields from documents	Date and amount from an invoice
Search	Answers from company documents	HR policy help desk
Forecast	Predicts a number	Next month's demand

A step that fits none of these patterns is rarely a first candidate.

Scoring Value

Value means the benefit if the use case works. Estimate it in the unit that your organization already tracks. For time savings, multiply how often the task happens, how long it takes, and how much of that time the tool can plausibly save. A task done 400 times a month for 10 minutes each is about 67 hours a month. If the tool halves the time, the saving is about 33 hours, which you can convert to cost using an average hourly cost.

Time is not the only form of value. Some use cases raise revenue, such as faster replies to sales leads. Some reduce risk, such as catching contract clauses that break policy. Some raise quality, such as more consistent answers to customers. For the matrix, convert your estimate to a score from 1 to 5 using ranges agreed in advance. Write the assumption behind each number so that others can challenge it. Do not claim precision you do not have.

Worked Example: Monthly Hours Saved

hours

Illustrative arithmetic from the text: 400 items a month at 10 minutes each is about 67 hours.

Scoring Feasibility

Feasibility means how likely you are to deliver the benefit in a reasonable time. Four questions cover most of it. Data: does the information the tool needs exist, in a form it can use, and may you use it? Technology: can it connect to the systems where the work happens? Risk: what happens if the output is wrong, and who checks it? Adoption: will the team use it, and does a business owner want it?

Score each question from 1 to 5 and average them, or take the lowest, since one weak area can stop a project. Data and risk are the most common blockers. You will study data in week 3 and compliance in week 5, so for now give an honest first score and mark the doubts. A use case with high value and a blocked data source is not a quick win, however attractive it looks.

The Value And Feasibility Matrix

Place each use case on a grid with value on the vertical axis and feasibility on the horizontal axis. Four zones appear. High value and high feasibility is a quick win: do it first. High value and low feasibility is a bet: it needs investment, such as better data, so plan it as a longer project with clear checkpoints. Low value and high feasibility is a fill-in: do it only if it costs almost nothing, or if it teaches the team. Low value and low feasibility should be dropped.

Aim for a portfolio, not a single pick. A healthy first list holds two or three quick wins that show results within weeks, and one or two bets that move the business over a year. Quick wins build credibility and skill. Bets create the larger change. Companies that only chase bets lose patience, and companies that only do quick wins stay small.

Examples And The One-Page Brief

Examples across departments show the range. Finance: extract fields from supplier invoices to cut manual entry. HR: draft answers to common policy questions for an internal help desk. Customer service: summarize a long ticket history before an agent replies. Legal: highlight unusual clauses in standard contracts. Sales: prepare a short account brief before each meeting. Software: generate tests for existing code.

Before any project starts, write a one-page brief. It names the process and the team, the problem in one sentence, the pattern (draft, classify, extract and so on), the baseline you will measure, the value estimate with its assumptions, the feasibility scores with doubts, the risk and who checks output, and the business owner. A manager should be able to approve or reject the idea from this page alone. Week 6 turns the approved brief into a pilot design.

Be ready for disagreement. A sales leader may rate the value of a use case higher than the people who do the work. Treat that gap as useful information and test it with a small sample or a conversation on the shop floor. Agreed scores, even if imperfect, are better than loud opinions.

Key Terms

Use case: A specific task in a specific team with a measurable goal where AI could help.
Process map: A written step-by-step picture of how work moves from trigger to result.
Trigger: The event that starts a process, such as an invoice arriving.
Handoff: A point where work passes from one person or system to another and often waits.
Rework: Work that returns for correction because something was missing or wrong.
Value score: A 1 to 5 rating of the expected benefit in time, revenue, risk or quality.
Feasibility score: A 1 to 5 rating of how likely delivery is, based on data, technology, risk and adoption.
Quick win: A use case with high value and high feasibility that shows results within weeks.
Bet: A high value use case that needs investment before it becomes feasible.

Exercise: Build Your First Ranked List

Choose one process from your team that feels slow or repetitive and write its trigger and result at the top of a page.
List every step with owner, system, time per item and common errors. Ask one colleague to check it.
Label each step with an AI pattern (draft, summarize, classify, extract, search, forecast, flag) or write 'none'.
Write three to five use case ideas from the labeled steps, each as one sentence with a team and a goal.
Score each idea for value and feasibility from 1 to 5, with one line of assumption per score.
Plot the ideas on a two-by-two grid in a spreadsheet or on paper and mark quick wins and bets.
Write a one-page brief for your top quick win using the fields from the document.