What Real Adoption From Top Lawyers Is Teaching Us

Real adoption from top lawyers has been the goal since day one.

Not a signed pilot that sits unused. Not a partner who opens it once, nods, and goes back to old habits. Attorneys at leading firms running their actual day out of Clevermore — that's the bar, and it's the only bar that matters.

If you haven't seen it: Clevermore lives inside Outlook and turns a lawyer's inbox into a working task list. It reads every thread, pulls out what actually needs doing, tracks where each client request stands, and files mail by client and matter along the way. No new app to open, no new workflow to learn. It works where lawyers already work.

Clevermore is now live inside our first firms, from AmLaw 100 to AmLaw 30. We went in eyes wide open: some of our defaults were always going to need to change once real mail hit them. You only learn certain lessons when the mailbox belongs to a partner carrying 60,000 items and a deal closing Friday.

Here's what real adoption across several firms has taught us so far.

We Found Something More Powerful Than Time Saved

Of course Clevermore saves lawyers time. That was the obvious promise going in, and it holds. What real usage revealed was a second effect we didn't fully expect — and it's bigger.

Within a day of going live, a director at an AmLaw 100 firm told us he hadn't realized how many open threads he was carrying until Clevermore captured them automatically. Seeing them all in one place was a genuine weight off his shoulders. A corporate partner at another firm put it better than our own marketing does: turn it on, do nothing, and it builds a living dashboard of every client request and where each one stands.

That's the discovery. The mental load a partner carries — the low background hum of "what am I forgetting" — is heavier than the minutes. Lift it and you've given them the certainty that nothing is slipping. For a partner whose reputation rides on responsiveness, that matters more than any productivity stat we could put on a slide.

But surfacing every open item creates a problem of its own.

Lesson 1: The Failure Mode Is Surfacing Too Many Things, Not Missing Them

Most teams building an inbox AI would obsess over recall — did we catch everything important. That's the intuitive worry. It's also the wrong one, and catching that early is a real edge. Lawyers sit on enormous numbers of threads they're only cc'd on. Turn every one into a task and you haven't built a to-do list. You've built a second inbox.

So we watched task volume from day one. A hundred-plus open items is noise, no matter how accurate each one is — and left unchecked, that pile turns a product people love into one more thing they avoid. We got ahead of it before it became a problem for anyone.

The first fixes came fast: opt-in auto-completion, so finished work clears itself. Controls to ignore a thread or an entire sender, so the cold vendor pitch never becomes a task in the first place. And there's a wide universe of approaches to this problem — surfacing one task at a time, sharper prioritization, smarter defaults about what deserves attention at all — that we're testing side by side with top lawyers right now.

Deciding what to leave out has become even more important than making sure nothing gets missed.

Lesson 2: Match the Way Lawyers Think, Don't Bring a New Framework

We shipped a category called "Junk" as a sensible starting default. A firm CIO sharpened it for us within days — correctly, it was too opinionated, sweeping up event invitations and marketing mail he actually cared about.

Good signal, and it pointed at something bigger than one label. Attorneys who've practiced for twenty years have elaborate personal systems for their mail. Folders by client. Folders by matter. Rules they've tuned since before we founded the company. Impose a new taxonomy on top of that and you're not shipping a feature. You're picking a fight.

So we made meeting the firm where it already works a core principle, and shipped to match. We retired the opinionated labels for a neutral set. We taught Clevermore to read a user's existing folders and file into them — their scheme, not ours. And we added a clean opt-out of auto-categorization entirely.

One detail shows how careful this has to be: Outlook applies categories at the account level, so a tag added on desktop shows up on the partner's phone.

Now our framework matches the way partners think, instead of asking them to adapt to something new.

Lesson 3: Security and Reliability Matter More Than New Features

The thing that makes Clevermore valuable — reading the whole inbox — is exactly what makes a firm's security team stop you at the door. The first question was never about features. It was about scope.

One firm's enterprise team wouldn't start the pilot until access was provably limited to the intended users and nothing else. After go-live, their engineers kept auditing it, and they exercised their right to wipe the environment clean and start over more than once. That's not friction to route around. That's the buyer telling you what the product has to be.

Our answer is architectural, and it's simple to state: we deploy into the firm's own Azure tenant, strictly isolated, with zero data retention — nothing stored, nothing trained on.

Reliability earns trust the same way. An Outlook add-in lives inside an environment firm IT manages with real rigor. One firm paused our update — not over anything we did — because a Microsoft patch had just broken their document-management add-in, and they weren't touching another moving part until it was fixed. Exactly the right instinct, and we respect the house rules that come with it. Smaller edges matter just as much: a pane that won't stay pinned reads as "broken" to a partner deciding in week one whether we earn a permanent spot on their screen. We clear those fast. For the IT director who has to answer for us internally, reliability is the feature.

What We're Taking From This

Several firms in, the pattern holds. The hard problems aren't in the model — they're in fidelity to how lawyers already work: what not to surface, whose framework to match, what a security team needs to see.

You earn a firm's trust by treating its constraints as the spec, not the obstacle. The firms letting us into their partners' mailboxes are telling us exactly what the product has to be — and the edge isn't hearing it, it's turning it into shipped product faster than anyone else. We're just getting started.