Building Your Use-Case Inventory
Use this worksheet to list the repeated work your practice actually does, then rank which tasks are worth carrying into the rest of The Adoption Path first. It is the companion to Decide What AI Is For, which explains why the list comes before the tool; this page is where you write it down.
What this measures
This worksheet measures two things about each candidate use and nothing else: how much repeated work the task represents, and how exposed that work is — what information it needs, who performs it, and whether a client ever sees the output.
Together they give you a score that puts your candidates in order. You end up with a use-case list that has a sequence to it, so the next steps start somewhere obvious instead of at eight tasks of equal apparent weight.
Before you start
Work from the last quarter of real work rather than from memory. Open your calendar and your sent mail and read them as a reminder of what actually recurs.
Two ground rules before you fill anything in:
- Include your assistant’s work. If an assistant drafts, files, schedules, or summarizes anything, those tasks belong on the list under that person’s name. A use nobody wrote down is shadow IT waiting to happen.
- Do not pick a tool yet. The inventory describes work, not software. If the firm already runs an advisor-specific platform, list the tasks you do inside it like any other; what changes is the second table, where your agreement with that vendor answers part of the data question for you.
Your recurring tasks
List the work the practice repeats. The test is simple: is this something the firm already does regularly, on a weekly or quarterly cycle? Aspirational uses — the thing you read about and would like to try — stay off, because you cannot estimate what they save or what they touch.
| Task | How often | Who does it |
|---|---|---|
| Draft the quarterly client letter | Quarterly | Adviser |
| Summarize a meeting before the next one | Weekly | Assistant |
| Search old notes for a fact | Weekly | Adviser |
| Draft a first reply to a routine client email | Daily | Assistant |
Write each task in the words you would use with a colleague. “Draft the quarterly client letter” is a task; “content generation” is a category, and a category cannot be scored or supervised.
What each task touches
For every task on the first list, answer three questions: what information it needs, who performs it, and whether a client sees what comes out. This is what the later steps work from, so it decides most of the score.
| Task | What information it needs | Who sees the output |
|---|---|---|
| Quarterly client letter | Holdings and performance by household | Clients |
| Meeting summary | Full notes, names, account detail | Adviser only |
| Search old notes | Note text; a query needs no client name | Adviser only |
| First reply to a routine email | The client’s message and file | Client, after review |
Be precise here. A summary that names nobody still carries balances and circumstances, and anything identifying a client on its own or in combination is client-identifying — the answer is not “none.” Where a client sees the output, note whether people who are not clients would see it too, because a communication aimed at prospects may be an advertisement with rules of its own. In a general-purpose tool rather than something built for advisory firms, treat the information question as mattering more, not less.
The result
Score each task on three questions, 1 to 3 on each, and add them:
- How often — 3 for weekly or more, 2 for monthly or quarterly, 1 for a few times a year.
- What it touches — 3 for no client information at all, 2 for client information you can strip out before the task, 1 for client-identifying information the task cannot work without.
- Who sees the output — 3 for you or your assistant only, 2 for another professional, 1 for a client or a prospect.
A total of 9 is frequent, low-exposure work that stays inside the firm; a total of 3 is rare, client-facing work built on information you cannot remove. Rank from the highest total down and keep the top few.
| Candidate use | Score | Where it ranks |
|---|---|---|
| Search old notes for a fact | 8 | First |
| Meeting summary before the next meeting | 7 | Second |
| First reply to a routine client email | 5 | Third |
| Quarterly client letter | 4 | Last |
What the number does and does not tell you
The score ranks candidates for the next steps. It does not clear any of them. A task scoring 9 is not approved, permitted, or vetted — it is only the first one worth carrying through Map the Rules That Apply and Sort Your Data. A low score is not a prohibition either; it means more work comes first.
Two further limits. The score says nothing about whether any tool does the task well, which is a vetting question. And it says nothing about what you owe the client: the recommendation is yours whatever drafted it — that is what fiduciary duty means here, and no total moves it.
What to do with it
Date the inventory, keep it with the firm’s other working files, and take the top item through the next steps rather than all of them at once. The list feeds those steps twice over: the information column becomes the start of the data matrix in Sort Your Data, and the ranked order tells you which use the written policy has to cover first.
Keep each dated version rather than overwriting the last. For an SEC-registered adviser, how often the compliance program must be reviewed is set by rule — see this year’s figures — so a dated series gives that review something to read; a state-registered adviser checks whether its state sets a comparable review, and the reason to keep the versions is the same. Where what a use produces is a record the firm must keep, how long it keeps it is set by rule, not preference — see Records and Documentation in this year’s figures.
Revisit it when the work changes, not on a cycle for its own sake: a new service line, a new staff member, or a task that quietly moved from quarterly to weekly all change the ranking.
At a larger firm
At a larger firm the first table usually exists already, kept by compliance or by whoever owns technology, so your job is to find it rather than start one. Expect the scoring to be somebody else’s scale, applied firm-wide; use theirs and do not invent a competing one. The second table does not change: for each task you want to bring, you still have to say what information it needs and who sees the output. Approval for a use elsewhere is not approval for your version of it, because the data and the reviewer may both differ.