Billables AI

Here is a fact about lawyers that is both obvious and, when you sit with it, faintly absurd: they sell time. Not documents, not advice, not outcomes - time, sliced into six-minute units and logged against a matter number. It is the oldest business model in the profession, and it depends entirely on the seller remembering what they sold.

They mostly don't. You answer a client email at nine at night. You take a call from the car. You mark up a contract on the train, spend twenty minutes on the phone with opposing counsel, and by Friday afternoon, when the timesheet is due, a meaningful chunk of that week has quietly evaporated. Not because the work didn't happen. Because nobody wrote it down.

Billables AI, a San Francisco company founded in 2023, has organized itself entirely around this gap. The premise is almost aggressively unglamorous. There is no shortage of legal-AI startups promising to draft your briefs or read your contracts or replace your junior associates. Billables picked the least sexy problem on the list - filling out the timesheet - on the theory that everyone has it, everyone hates it, and it maps directly onto money.

And it does map onto money. The company's headline claim is that firms using it capture 10 to 30 percent more billable time. Sit with that range. It means that a nontrivial fraction of the work these firms already do is simply never invoiced. The lawyers did it. The clients arguably owe for it. It just fell into the crack between doing the work and recording it, and Billables' entire product is a bet that the crack is worth a company.

The way it closes the gap is the interesting part. Billables runs quietly in the background and reconstructs your day from the digital exhaust you generate anyway - the email you sent, the calendar block, the call that showed up in the log, the document you had open. Then it does the thing lawyers hate most: it writes the narrative, the little description that says what you did and why it was billable, formatted the way the client wants to see it.

You review, you edit, you approve. The machine has already done the remembering and the writing. The human just signs off.

There is an easy way to build this product and a hard way, and Billables chose the hard one on purpose. The easy way is to watch the screen - record keystrokes, take screenshots, track every window. Plenty of monitoring tools do exactly that, and lawyers, being professionally paranoid about privilege and privacy, tend to hate them.

Billables instead reads metadata through secure API integrations with the tools a firm already pays for: Microsoft 365, Google Workspace, Zoom, Adobe, and practice-management systems like Clio, MyCase, LeanLaw, SurePoint, Centerbase, and Litify. It knows a call happened and who was on it. It does not need to know what was said. Same signal, none of the surveillance.

This is a genuine product distinction dressed up as a privacy feature, or maybe the other way around. The creepy version and the useful version of this tool start from nearly identical data. The difference is entirely where you draw the line, and Billables drew it somewhere lawyers can live with. That, more than any model architecture, is probably why hundreds of them signed up.

It also explains the integration list, which reads like a chore but functions as a moat. The flashiest AI in the world loses to the unremarkable tool that already lives inside the workflow. Billables doesn't ask lawyers to change how they work. It just watches - politely, at the metadata layer - and turns the watching into invoices.

The founding team is more decorated than the problem might suggest. Chief executive Arvind Sujeeth previously built enterprise AI platforms at SambaNova Systems, the well-funded AI-chip company - which is to say he arrived at legal billing from the deepest end of the machine-learning pool. Co-founder Nancy Jeng ran global product marketing at Pinterest. Laura Maddox, the founding VP of engineering, built conversational AI at Mya Systems.

It is a specific kind of team: people who have shipped hard technical products to large audiences and then, apparently, looked around for a problem where all of that could compound. Legal timekeeping is a strange and excellent choice. It is universal within its market, structurally tied to revenue, and - critically - a task humans are genuinely bad at. Remembering and describing your own work is exactly the thing machines can do better. Build the company around that asymmetry and you don't need a large team.

Which is convenient, because the company is small. Around a dozen people served those hundreds of firms as of its early growth. That ratio is only possible if the software is doing the labor, which is both the pitch and the proof.

Billables emerged from stealth in October 2024 with $3.9 million in seed funding led by Wing VC, alongside F7, SignalFire, Darkmode, Alumni Ventures, and a roster of angels drawn from Google, Stanford, and SambaNova. Standard-issue promising-seed-company stuff.

Then, in June 2026, it raised a $10.2 million Series A led by Avenue Growth Partners, with Wing, SignalFire, and Alumni Ventures returning. And somewhere in the gap between those two rounds, the story quietly changed. The seed pitch was about timekeeping. The Series A pitch was about "operational intelligence." Same product, considerably bigger frame.

You can see the logic. Once you are capturing every timekeeper's activity in order to bill it, you are also, incidentally, capturing how the entire firm runs - who works on what, where the hours go, which matters bleed margin. Timekeeping was the wedge. The data underneath it is the business. Avenue's Brian Goldsmith put it plainly: the company had established itself in AI-powered timekeeping, "but its broader opportunity extends far beyond billing."

Billables is not alone in noticing that timesheets are miserable. It competes with a mix of AI-timekeeping upstarts and incumbents - Ping, Laurel, WiseTime, Aderant's iTimekeep, and the manual timers baked into practice-management suites like Smokeball. The category has existed long enough to have a graveyard.

Its bet against that field is the same one running through everything else here: privacy-protective, API-based capture that respects how lawyers already work, sold into the midmarket where firms are large enough to bleed real money on unbilled time but small enough to still feel it. Whether "operational intelligence" is a category or a fundraising deck is the open question. For now, the timekeeping works, the firms are paying, and the award is on the shelf.