Get your intake hours back.

The ABA's Standing Committee on Ethics issued Formal Opinion 512 on July 29, 2024. It is the first formal ABA guidance on generative AI, and it is short enough to read in one sitting. For a small estate practice it comes down to four things.

1. Know what the tool does (Rule 1.1). You do not need to understand the model. You need to know what goes in, what comes out, and where it fails. Our handoff includes a one-page system map for exactly this reason.
2. Confidentiality and consent (Rule 1.6). Client information entered into a tool that learns from it, or that a vendor can read, needs the client's informed consent. We only use Claude through the API on your firm's own account, which is not used for model training, and we document the data path so you can decide what, if anything, to tell clients.
3. Tell clients how the work gets done (Rule 1.4). The opinion ties AI use to the duty to consult clients about how you pursue their objectives. Many firms add one sentence to the engagement letter. We can suggest a starting draft for you to review, but the wording is yours.
4. Bill for the review, not the time you saved (Rule 1.5). The opinion's own example: a lawyer may bill the 15 minutes spent entering facts and the time spent reviewing the draft, but cannot bill the hours a draft would have taken by hand. A flat-fee practice sidesteps most of this, which is one reason Sable Law suits estate work.

None of this is legal advice, and your state bar may have added guidance since. Read the ABA's announcement and check your state bar before you decide anything.

The first hour of any build is not building. It is reading how the firm already works, because every automation inherits the habits underneath it. Here is the list we work through, in order.

1. Who owns the accounts. The Clio subscription, the Google Workspace admin and the Lawmatics login should belong to the firm, not to an employee or to us. If a departed paralegal still owns the Workspace, we fix that before anything else.
2. How matters are numbered and named. We match whatever convention the firm uses today. A workflow that invents its own numbering breaks the filing cabinet.
3. Which custom fields are real. Most Clio accounts have fields nobody fills in. We find the five or six that staff actually use, and the workflow writes only to those.
4. Where the templates live and who last edited them. If the engagement letter exists in three versions, we pick one with the attorney and retire the others before the workflow fills it.
5. Who approves what. We write down, in plain words, which outputs the paralegal can release and which need the attorney. That list becomes the approval step inside the workflow.
6. Least access. Each integration gets only the permissions its job needs, and credentials stay in the firm's own accounts. We never hold a master password.
7. How to turn it off. Before go-live, the paralegal switches the workflow off and back on without us. If she cannot, we are not done.

Most of this is decided on the scoping call. It is also why the memo you get afterwards can include a fixed quote: by then, there are no surprises left in the account.

The savings calculator on this page uses example inputs. The number that matters is yours, and we get it in about ten minutes at the start of the call.

1. Pick one real intake from the last month. A routine one, not the hardest. Sample or redacted data is fine.
2. Walk it from first contact to opened matter, naming each person who touched it and each place the same facts were typed: the intake form, Clio, the spreadsheet, the engagement letter, the welcome email.
3. Count the re-keying. The number of times the client's name, matter type or fiduciaries were typed by hand is usually the whole story. Four or five is typical.
4. Add the waiting. Days between consultation and signed engagement letter, and how the firm found out when a client went quiet.
5. Put a time on it, using the paralegal's estimate, not ours. We write that number into the memo as the baseline, and the same intake is timed again after the build.

That before-and-after figure is the only number we quote back to you. If the baseline turns out to be small, we say so on the call and you keep the memo.