Services / Mail automation
Mail that files itself the moment it's scanned.
We build custom mail automation for law firms. A scanned letter gets read, sorted into the right folder, matched to its case, and logged before anyone on your team has opened the PDF. The letters that carry real risk don't get auto-filed at all. Those go to a person.
The problem
Scanning solved the paper problem, not the work problem. Someone still has to open each PDF, figure out what it is, find which client it belongs to, and put it where it goes. At any real volume that's a queue a paralegal feeds hours into every week, and it's exactly the kind of repetitive reading where a tired human misfiles the one letter that had a deadline in it.
The uncomfortable part is that the dangerous letters look almost identical to the routine ones. A decision letter and a records receipt arrive in the same folder, in the same scan batch, and the cost of getting them confused is not symmetrical.
How we build it
Read and classify
AI reads each newly scanned PDF and classifies it into your firm's own folder taxonomy. At the firm this runs for, that's 17 letter types, and classification takes under a second per file. The Zapier chain it replaced took minutes.
Match and file
The letter gets matched to the right Clio matter, with fuzzy matching that survives the name typos scanners introduce. It's filed to its folder, duplicated into the month's archive, and the matter gets updated, all in one pass.
Gate the risky mail
Decisions, Appeals Council letters, late filings, and good-cause letters never auto-file. They land in a review folder and a person gets emailed. At the firm this runs for, that's roughly 4 letters a day pulled out of the flow on purpose.
Log everything, drop nothing
Every file writes an audit row. A failure keeps the letter in the inbox for retry instead of losing it, and an hourly safety net re-drains anything the real-time trigger missed.
Where it runs today
This runs live at a Social Security disability firm on every piece of mail it scans. A second stage files worked mail into each client's folder too, and when a match is genuinely ambiguous the system emails a one-click question to a human instead of guessing.
17
letter types sorted automatically
< 1s
to classify a letter
~4/day
high-stakes letters gated to a person
Common questions
What happens when a letter is ambiguous?
It doesn't guess. An ambiguous match generates a one-click email to a person who picks the right case, and high-stakes letter types skip automation entirely and go to a review folder. The system is built on the idea that a wrong guess on legal mail costs more than a short wait for a human.
Do we have to change how we scan?
No. It watches the folder your scans already land in and works from there. Your scanner, copier, or mailroom process stays exactly as it is.
What if it files something wrong?
Every action is logged, so you can see exactly what went where and why, and move it. Failures are loud: the file stays in the inbox and we get alerted. Nothing disappears silently, which is the failure mode that actually hurts.
Buried in scanned mail?
Book a free audit and we’ll walk your setup and tell you what this would look like at your firm, including what it would save.