Over the past two years, most large firms have run pilots, compared platforms, and chosen the AI tools their litigators will rely on for years to come. The next question is quieter, but it decides how much value those tools deliver: when your lawyers log in, whose court data is the AI actually reading?
For many firms, it is a mix: someone else's case law, licensed from a legal publisher, plus their own filings, often saved by hand into the DMS and not always in full. What is usually missing is the complete trial record: every filing on their cases from complaint to final order, captured as it posts and tagged with its client and matter codes.
We have spent more than a decade capturing court filings for law firms, and now that AI reads those records, the question that matters most, and that many firms haven’t asked yet, is whether a firm holds an accurate, complete, and timely record of its own litigation.
What "court data" covers
Court data is everything the court system generates on a case: the docket sheet and its entries, the notice sent when something is filed, the filing itself and the metadata that describes it. AI can read all of it. What court data can't tell AI is which client and matter a filing belongs to, because that context lives inside your firm, not on the public record.
The docket: the court's description of each filing
The docket sheet is the court's running list of a case, and each docket entry describes an underlying filing. The entry stays on the docket sheet, and in federal court counsel are also notified in real time through a separate Notice of Electronic Filing (NEF), which carries the docket text and a link to the document (PACER FAQ). For an AI tool, that is a headline without the story: it learns that a motion was filed, but nothing about what the motion argues.
The document: the filing itself
The document is the filing itself, where the arguments, the facts and the court's reasoning live. Saved as a PDF, it is harder to find, file and trust, because the content is there but the structure around it is missing.
The metadata: the details that describe each filing
Each filing comes with metadata: the judge, nature of suit, filer, filing type, filing date, counsel, jurisdiction, case filing date and docket description, among more than 50 fields. It is what lets a system, or an AI tool, tell one filing from the next.
Your client and matter: the context court data doesn't carry
The client matter isn't court data. It's your firm's confidential context, and it's what makes the public record actionable: the matter workspace ties each filing to the internal memos, legal research and correspondence behind the case, and to what the exposure actually was.
The two ways firms get their court data today
Without automated capture, firms collect the court data on their cases by hand, rely on a legal publisher, or do both, and each route leaves gaps that are hard to see from the partner's chair.
Manually: someone downloads it, which can be inconsistent and sacrifices critical metadata
When a filing posts, someone on the case team clicks the link in the NEF, saves the PDF, names it and files it, provided they handle that matter and are at their desk. Docketing teams handle this at a volume most partners never see: Husch Blackwell received 78,836 unique NEFs, an estimated 400,000 total NEFs, and managed 103,502 filings in a single year.
But a process that depends on who clicked produces a record that depends on who clicked. File names vary, metadata is lost because there is too much of it to retype on every filing, and filings that land after hours are the easiest to miss, so even firms with large central docketing teams capture 25% more filings after deploying PacerPro.
From a publisher: possibly your data, incomplete, without your firm's information, licensed back in incomplete form
Publishers solve a different problem, and they solve it well. Their core is published case law, and many also carry dockets and selected trial court filings, some of which may be yours. But a publisher holds a small cross-section of everyone's data, not all of yours, without your client codes, and you read it under its license, so parts of your own litigation end up being licensed back to you.
Either way, you're left with an incomplete record and incomplete metadata.
The third model: a workflow tool delivers your filings for you
The third model is a tool embedded in your own workflow, which ensures each available filing on your cases is delivered into the DMS as it posts, profiled with its metadata and your client and matter codes, under the same rules on every matter. It is the model behind PacerPro's Manifold.
Why "outside vendor" sounds like "third-party data"
The concern is reasonable, and we'd rather address it head-on. The word doesn't help, since a publisher is a vendor too, so the real difference lies in where each one sits.
A publisher sits outside the firm and builds one database for all its subscribers, while a workflow tool sits inside your workflow and covers only your own cases, filing them under your matter structure and following your rules for naming and access. Any vendor still deserves security diligence, and PacerPro is SOC 2 Type 2 certified, with its controls published in its Trust Center.
What lands in your DMS: documents, docket sheets, metadata, parties
For IT and knowledge teams, the practical question is what lands in the DMS. With each filing, you receive the document and its available attachments and exhibits, consistently named, along with its docket sheet, 50+ metadata fields, the parties and their roles, and your own client and matter codes. Over time, that builds into the record of a single matter or of your whole litigation history.
What publishers hold, and what PacerPro delivers
Publishers hold published case law, most of it appellate, which tells you what the law is. PacerPro delivers your own trial court record into your DMS, which shows how that law was applied to the facts.
Publishers: appellate level. What the law is.
Brown v. Board of Education is still cited for the principle it established, which is the job of published case law: settling the rule that later cases apply to their own facts. Reported cases come mostly from appellate courts (UW Law Library), and not every appellate decision is published, but every litigator relies on this layer.
PacerPro: your firm's trial record, with your codes. How the law was applied to the facts.
The cases that applied Brown were fought out in trial courts, where lawyers argued over a particular district's zoning, classrooms or facilities through complaints, motions, exhibits and testimony, and a judge weighed the evidence. That is where facts are found, and the back-and-forth reveals as much as the final ruling.
On the defense side, where large firms do much of their work, a $20 million outcome on a $200 million exposure can look like a loss on the docket sheet while counting as a success inside the firm. PacerPro captures that whole story, from complaint to final order, coded to your clients and matters.
Why you need both, and why procurement often only knows the first
Attorneys grasp this immediately, but procurement’s first reaction is often “we already get court data from our research provider”, when what most firms buy there is published case law, which no litigator can do without, plus dockets on a selection of cases. The trial record is the other half: the operation rather than the theory. A more useful question at renewal is whether a source covers every case, client code and jurisdiction you work in, and stays current as filings post.
Why your court data decides what your AI can do
An AI tool can only reason over what it reads, so on published case law it knows the law, and on your captured record, it also knows your own cases. In ILTA's 2026 Technology Survey of 508 law firms, 94% said they were using or exploring generative AI, up from 80% a year earlier, and among firms with 700 or more lawyers, the figure was 100% (ILTA 2026 Technology Survey). What sets firms apart now is how they use those platforms.
On a publisher's corpus, it knows the law
On published case law, an AI tool finds the controlling authority and traces how a rule developed. Ask it what you argued in one of your own cases last year, though, and it has fragments at best, with no matter code to find them by.
On your captured record, it knows what your firm filed, where, and what came before
Connected to your own trial record, filed by matter, the same tool can tell a partner on Monday morning what opposing counsel filed over the weekend, with the answer pointing to a filing in the right matter. Its answers are only as good as the record, though. If exhibits are missing, the AI may assume a winning motion rested on strong evidence when the briefing was carrying a thin record, and nothing in its answer will flag the gap.
Our Data Teams page shows how this data is structured for this use case.
Where PacerPro sits
PacerPro sits between the courts and your systems, profiling and delivering the filings on your federal and state cases into your DMS as they post, named to your own conventions and structured for AI to read.
For your AI, this is the difference between reasoning over fragments and reasoning over the case. Whichever platform you have chosen, PacerPro gives it a record of your own cases, with each filing in the right matter and described by its metadata, so a partner’s answers rest on the case as it was actually litigated. For firms working with Harvey, PacerPro's Harvey integration syncs a case's full docket into a Harvey Vault and keeps it current as new filings arrive, so litigators can query, summarize and draft against the live record.
The trial court data layer for law firms
PacerPro is the trial court data layer for law firms. Publishers remain the source for what the law is; we supply the record of how your own cases unfolded.
Eight of the ten best US firms; most of the Am Law 100
The majority of the Am Law 100, including eight of the ten best US firms, rely on PacerPro for their court data. As firms settle on their AI platforms, the ones that hold a complete record of their own litigation will get more from whichever platform they choose.
See how it works on your matters. We'll set up a pilot on a handful of your active matters.
FAQ
What is court data for a law firm?
Court data covers everything the court system generates on a case: the docket sheet and its entries, the notices sent when something is filed, the filings themselves and their metadata. Your client and matter codes aren't court data. They're your firm's own context, and they're what tells AI which matter a filing belongs to.
What is the difference between appellate opinions and trial court records?
Appellate opinions are published decisions that set precedent and tell you what the law is. Trial court records are the motions, rulings and exhibits in a case, which show how the law was applied to the facts, and most are never published.
Do legal publishers hold my firm's court filings?
Possibly some, since many publishers carry dockets and selected trial court filings, but they won't hold every filing on your cases, and none of it carries your client and matter codes.
Is a vendor-captured court record third-party data?
Not in the usual sense. A capture tool embedded in your workflow collects only your own filings and delivers them into your DMS, under your own matters and rules.
Who captures trial court records for law firm AI?
PacerPro, the trial court data layer for law firms, captures federal (PACER and CM/ECF) and state filings on a firm's cases as they post and delivers them, with metadata and client and matter codes, into iManage or NetDocuments.


