How does Ajax figure out which client and matter each activity belongs to?
Ajax figures out the client and matter behind every activity by reading what's on your screen and matching it against a live map of your firm.
Most of the time, the client's name or matter number is right there in the document or email, and Ajax makes the match directly. When it isn't, Ajax leans on a knowledge graph that gets populated from your billing system and past communications before anyone at your firm ever logs in.
Every entry you touch after that helps Ajax become more accurate - unchanged entries strengthen its guesses, changed ones weaken them. And a semantic layer on top links related activity across different apps, so even work with no useful contacts on screen still gets attributed to the right matter.
Let's get into the details of how each aspect of this works.
How does Ajax see what you're working on in the first place?
Ajax runs on your device and takes screenshots of your screen, then reads the contents. Whatever's on the page in front of you throughout the day (the documents, the emails, the browser tabs, the contents of every file as you scroll through it) is what Ajax is capturing.
This is a broader set of information than any other timekeeping tool works from, which is why Ajax is able to be so effective. Most tools rely on calendar entries, app titles, or after-the-fact prompts to guess what you did. Ajax sees what you actually did throughout the day, then uses that to pick the client and matter for every entry it generates.
That's what makes the whole matching process possible. A tool that only knows you had Word open at 2 p.m. can't tell you which brief you were drafting. A tool that sees the contents of your screen can.
An important note on data privacy and security
The is the number one question every law firm asks us. The short version: Ajax doesn't hold onto your data, doesn't train on it, and nobody at your firm can see what's in your Ajax except you.
Here are the key things to know about how we handle your data:
Rolling deletion. Screen contents are used to generate your time entries and then deleted on a rolling basis
No model training. Your firm's data is never used to train Ajax's models, or anyone else's
No vendor retention. The AI providers Ajax uses are contractually prohibited from retaining or training on your data
SOC 2. Ajax is SOC 2 compliant
Individual silos. What's in your Ajax stays in your Ajax. Managing partners can't see it. IT can't see it. It's yours
Now let's get into how the actual matching works.
How does Ajax usually connect an activity to the right matter?
Most of the time, the client's name or the matter number is sitting somewhere in the actual contents of your screen. It's in the email you're writing, the document you're editing, or the API response you're looking at. Ajax is able to see that and make the match.
Ajax knows the full list of your clients and matters because it's integrated with your billing system. When you first sync Ajax up, we pull in every client, contact, and matter, and that list is continually refreshed as your firm adds and closes work. So the list Ajax is checking against is always live, not a stale export from three months ago.
What if it's not so obvious which matter you're working on?
A lot of the time, the matter isn't obvious from what's on screen. For example:
If you're on a call with someone whose name doesn't appear in your billing system yet
You're taking notes in a scratch doc that doesn't reference the case
The email you're writing doesn't mention the matter number.
Ajax handles these cases with a knowledge graph. It's a running map of the context around your work: which people are associated with which companies, which names, which words and phrases show up in which matters.
We populate that graph before anyone at your firm actually uses Ajax. We parse the names of every matter in your billing system, look at your previous communications, and cross-reference the two. So if a contact has shown up in 20 past time narratives and 19 of them were for one thing, Ajax creates a rough weighting of who they're most strongly associated with. That happens day zero, before anyone at your firm has logged a single entry.
Which means on day one, Ajax already has a picture of your book of business specific enough to guess correctly on a lot of the activity that has no client name or matter number on screen at all.
How does Ajax get smarter about your firm over time?
The day-zero population is only the starting point. From day one onward, Ajax learns from you. Every entry you touch (or don't) teaches Ajax more about how to accurately match client work to the correct matters.
When you push an entry into your billing system unchanged, Ajax reads that as confirmation and strengthens the association between the contacts on that screen and that matter. When you change the matter first, Ajax reads that as a correction, weakens the old association, and often creates a new one with the matter you picked instead.
How much Ajax learns from a single touch depends on how important the contact was in what was on your screen. If it's the person you were emailing about a matter, Ajax learns a lot. If it's one of 15 ancillary witnesses mentioned in a filing, Ajax learns a little. The weighting isn't uniform, because not every contact carries the same weight.
How does Ajax understand the meaning behind your work?
Ajax also understands what your work is about. Most timekeeping tools group related activity by matching surface details: whether two activities used the same file, sat in the same folder, or had the same window title.
Ajax does that too, but it also reads what's inside the file.
So work on the same matter gets connected even when it's spread across different files, folders, and apps.
A concrete example
Let's say you look something up in Lexis, flit across a few cases, grab some text from one, and copy-paste it into a scratch doc. Then you draft a Word doc that doesn't use the exact text but references a case name from the scratch doc and from Lexis.
Ajax can say, okay, these three things are clearly related. They're all legal research with the same thrust of research and drafting intent, even though they happened in different apps.
That grouping is what lets Ajax attribute work even when the contacts on screen aren't useful.
Take that Lexis session on its own. You're not emailing anybody. The people whose names show up on the page are parties to precedent cases, not to your matter. Ajax has almost nothing to work with from the contacts alone. But the motion you're drafting in Word does have the actual parties to your case in it, so Ajax is very confident about which matter that motion belongs to. When Ajax links the motion to the Lexis session by meaning, the motion's confidence pulls the Lexis session in with it, and both get attributed to the right matter.
So when Ajax is confident about one piece of activity, everything linked to it by meaning goes to the same matter. That's how the Lexis session ends up on the right case, even though the names on the page don't tell you anything on their own.
Do I have to manually set up any of these relationships, or tell Ajax who's who in each case?
No. All of it happens automatically.
The preloading of your matters and contacts runs before your first day on Ajax. And then Ajax learns from your corrections once you start using the product. There's no relationship-mapping exercise, no CRM-style entity setup, no "please confirm this contact is associated with this matter" pop-ups.
If Ajax gets something wrong, Ajax will learn when you change the entry to correct the matter. And in practice users find that Ajax predicts some unknowns on day one and almost zero unknowns by day three, because it very quickly picks up exactly who's related to the matters they're working on.
Final thoughts
You don't have to teach Ajax who's who or map any relationships. Ajax handles all of that in the background: reading what's on your screen, matching it against your firm's clients and matters, filling in the gaps with what it already knows about your people, and getting sharper every time you fix an entry. You just do your work, and Ajax figures out the client and matter for you.
If you want to see how that plays out on your firm's actual work, book a demo. We'll walk through your firm's setup, show you what Ajax would look like on your matters, and scope a pilot if it seems like a good fit.
