The problem: the file that doesn't fit
Most files follow a pattern. Some don't. The contract, the account statements and the client's summary disagree with each other. A date is off. A balance doesn't reconcile. The question goes back to the client, and until the answer comes back, the file sits half-finished while the paralegal carries the open issues in their head.
These files take the most time, and they carry the most risk of a wrong number reaching a filed document.
The case: a default judgment declaration on a messy file
At a San Diego collections firm, a paralegal needed a declaration for a default judgment on one of these files. The system drafted it in about 15 minutes while she worked on other matters. Her review took 30 minutes. By hand, on a file this tangled, the same work runs well over an hour.
About 15 minutes of unattended drafting. 30 minutes of review. At least 30 minutes saved on a task that would have run over an hour.
The time saved is the smaller part of the result.
How does AI help when the documents contradict each other?
The draft came back with the problems already identified. It pointed to exactly where the source documents disagreed, so the paralegal knew what to ask before contacting the client.
The client sent back proof that settled the question. The paralegal gave that proof to the system. It did not start the declaration over. It revised only the passages that depended on the corrected facts, then reported what it had changed and why, in plain language.
"It gave me the response that pointed exactly where the differences were… I didn't have to worry about having to figure out how to word it or how to explain it."
Fixing a number is easy. Explaining a discrepancy in a sworn declaration, so that a judge reading it cold understands what happened, is where the time and the worry go.
Why are corrections so stressful?
Making a correction is quick. The worry is missing one.
A single fact on a case, whether a payment date, a balance or the spelling of a name, doesn't live in one place. It appears in the declaration, the summary of the case, the interest calculation, the proposed judgment and possibly a letter that already went out. Correcting it by hand means remembering every place it appears. Relying on memory is how an old number ends up in a filed document.
On a routine file, that's a nuisance. On a file where the client sends new information in pieces, it means checking the same document three times.
How does the system keep track of changes?
Every case has one running record: the key facts, where each one came from, and a dated note of every change and the reason for it. Documents are drafted from that record, not from anyone's memory of it.
When a fact changes, it is recorded once. The system finds the passages that relied on the old value, revises them, and reports what changed, where, and the source of the new information. The paralegal checks that short list instead of searching every document on the file. Months later, when someone asks why a balance differs from the client's first spreadsheet, the answer and the date are already on record.
The result: lower stress, earned trust
The paralegal started out skeptical. It was a new system, and she is the one who answers for a wrong number. A smooth, routine file didn't change her view. Watching the system take a correction on a messy file and get it right did. In her words, it has her back.
The work that needs a person stayed with her: noticing something was off, calling the client, judging whether the explanation held up. The follow-through, making sure the new facts landed everywhere they needed to, stopped being the slow and nerve-racking part.
What does this mean for the attorney?
Attorneys are responsible for supervising the people and the tools that prepare their filings.1 A record of what changed, when, why and on what evidence makes that supervision practical. The attorney sees the correction and the proof behind it in one place before signing.
A person still approves every document before it goes to a court or the other side. The system makes sure the document being approved matches the facts already confirmed.
What this case shows
- The messy file is where the gain showed. Here, the file with conflicting documents is where the system saved the most time and worry.
- Finding the discrepancy is half the work. A draft that flags where documents disagree turns a vague problem into a specific question for the client.
- Corrections should be recorded once and applied everywhere. One record per case, with a log of every change, means nothing depends on memory.
- Less stress is a real result. A paralegal who trusts the follow-through spends attention on judgment, not on re-checking.
Common questions
When a fact changes, does the AI rewrite the whole document?
No. The new fact is recorded once, and the system revises only the passages that relied on the old one, then lists exactly what it changed and why.
How does AI help when a client's documents contradict each other?
Before drafting, it compares the source documents and points to exactly where they disagree, so the paralegal knows what to ask the client.
Who approves the changes?
A person. The system drafts and reports what it changed; the paralegal checks it and the attorney signs.
Sources
- American Bar Association, Standing Committee on Ethics and Professional Responsibility. (2024, July 29). Formal Opinion 512: Generative Artificial Intelligence Tools (discussing supervisory duties under Model Rules 5.1 and 5.3). americanbar.org