When a timekeeping tool says it automatically matches emails and calls to the right client matter, that statement comes with an asterisk that is worth reading carefully.
We recently spoke with a partner at a small litigation firm who identified this asterisk in concrete terms. He liked the concept of passive timekeeping. He used Clio, his firm communicated primarily through Gmail and video calls, and the basic architecture of a communication-capturing tool made sense for his practice. But when he started asking specific questions about how matter matching actually worked, the limitations became clear enough that he decided to wait for a more mature version of the product before committing.
His reasoning was not a complaint. It was an honest assessment of what his firm needed versus what was available, and it offers a precise map of where legal timekeeping AI still has real work to do.
How Matter Matching Works, and Where It Breaks Down
Most timekeeping tools that automate matter matching use some version of the same approach. They look at who is in the email or the call, check those contacts against the related contacts listed in Clio for each matter, and use that association to assign the entry to the right file.
For this to work reliably, the contacts on a matter need to be populated in Clio. The primary client, obviously. But also the other parties who are likely to appear in communications: opposing counsel, co-counsel, expert witnesses, insurance adjusters, and anyone else who regularly sends or receives emails in connection with a specific case.
At a firm that focuses heavily on communication with clients and where matters are relatively clean, this works well. But at a litigation firm, the participant list on any given matter can be large and constantly changing. New parties get added. Opposing counsel changes. And critically, a significant portion of internal case communication happens between attorneys at the same firm, none of whom are listed as contacts in Clio.
He described a common scenario: a video call with just attorneys on both sides, no client present. That call is billable. But because the participants are all lawyers whose contact information is not necessarily linked to the right matter in Clio, the timekeeping tool cannot confidently infer which file the call belongs to.
He also noted that even if a workaround existed, such as adding opposing counsel to the related contacts field in Clio, that workaround itself would create a different problem. Opposing counsel does not belong in the related contacts field. That field has a specific meaning in the context of a legal matter, and populating it with adverse parties for the sake of a timekeeping integration introduces a kind of data corruption that careful firms rightly want to avoid.
The Profitability Question Behind Flat-Fee Work
One thing worth noting from this conversation is that this firm, while doing a significant amount of flat-fee work, still tracked all their time carefully.
The reason is profitability analysis. When you bill a client a flat fee for a matter, you still need to know how much time you actually spent on it. If a matter that was supposed to take twenty hours consistently runs to forty, you are losing money on your flat-fee arrangement whether or not you realize it. Tracking time on flat-fee matters is how firms figure out whether their pricing is sustainable.
This reframes the conversation about who needs timekeeping tools. It is not only firms billing hourly. Any firm that wants to make informed decisions about its pricing model, whether to continue taking flat-fee work, whether to adjust rates, whether to staff certain matter types differently, needs reliable time data. The billing is almost secondary. The management information is what drives the practice forward.
What a More Capable AI System Would Look Like
He was clear about what would have to change for a tool like Lawgbook to work well for his firm.
The core requirement was content-based matter inference. Rather than relying solely on participant matching against Clio contacts, the system would need to be able to read the substance of a communication and use that to identify the right matter. If an email thread has been about a specific case for months, the language in the thread carries information about which file it belongs to. A sufficiently capable AI could learn to recognize those patterns, especially with some initial calibration.
He also suggested that the tool should be able to assign matters based on domain matching for email. Most of his clients are companies with their own domains. If the firm receives email from a contact at that domain, the domain itself is a strong signal pointing to the relevant matter, even if the specific person's email address has not been manually added to Clio.
Both of these are reasonable expectations. Content-based matter inference is technically harder than contact-based matching, but it is the direction the better tools in this space are heading. Domain-based matching is more immediately achievable and would already resolve a significant portion of the gap he identified.
The Bottom Line for Litigation Firms
What this conversation illustrates is that the gap between what AI timekeeping tools promise and what they deliver is not uniform across practice types. For a solo attorney who sends emails primarily to a small, stable client list, matter matching via contact lookup probably works well most of the time. For a litigation firm with a large and dynamic cast of participants on every matter, it is likely to require manual correction on a meaningful percentage of entries, which erodes the value of the automation.
The way forward is not to ask litigation firms to restructure how they manage contacts in Clio to fit the tool's limitations. The way forward is to build tools that can infer matter associations from context rather than only from contact lists. That is where the product needs to go, and this attorney's feedback is a clear and useful signal about why.