How an M&A advisor sends 446 replies he didn't write

Two weeks to get a set of financial statements

A diligence engagement runs four to eight weeks and lives entirely in email. The target is a small business, so its numbers are scattered, and the work becomes a chase.

“We kicked this one off two weeks ago. I've been spending two weeks just trying to get some basic financial statements. That's when the email gets real heavy — getting this, thanks for sending that, we still need this, and thank you for sending that but it's not all the way right, we really needed 2025 not 2024. And that can go on for months.”

Then the second cost, which is finding what already happened. “There was something I needed for a client that I know it was in an email. So I went to all mail and I had to search for it… that's an extra step I really don't want to take, because I'm still having to read everything.”

Answer the meeting request without the four-email dance

The first thing he asked us to fix was not diligence. It was the exchange that happens before any of the work starts.

“Somebody hits my inbox and says, we need to have a meeting. And I say, okay, when are you available? And what ends up happening is this back and forth. How about Tuesday?”

He has a booking link. He does not always use it, because the reply is a judgment call before it is a calendar entry. “Ideally it would be, here’s my link, you book. But sometimes it’s not.” An introduction from a trusted source gets a different answer from a cold pitch, and both arrive looking like a meeting request.

So the draft has to carry the judgment, not just the link. Our first attempt did not, and he said so: “It sounds very canned, which is fine, but it just doesn’t give it that.” What he wanted was the version he would have written: thanks for the thoughtful introduction, would love to learn what you’re working on, time below.

The availability behind it had to be real too. His personal calendar was joined to his work one so a proposed time is a time he can actually take.

Learn how he talks from how he replies

He asked the question most people never do, on the first call: is it learning from what I send, or am I supposed to teach it?

“As I’m responding to emails, is your machine learning about how I’m responding? Then eventually at some point it’s going to be like, how this guy talks.”

Both, is the answer. It reads what he sends, and anything he wants it to know for certain he can write down once. He was weighing which to lean on:

“Is it worth putting some stuff in there, or should I just let it do its thing and then one day look down and be like, yeah, this is how I would say it?”

The reason that matters in his business more than most: the difference between a good reply and a bad one is rarely information. It is register. He can tell instantly when a sentence is not his, which is why a draft that is 90% right still costs him the time to rewrite it.

He tried to build it himself first

Before Clarity's project view existed, Herman built his own with a chat assistant: a spreadsheet with a timeline per engagement, objectives, key activities and deliverables by week, toggled between complete, in progress and delayed.

“We just kind of created this and Claude… before I go to ClickUp and HubSpot and do all that kind of stuff. I'm trying to keep it low tech around here, and low information diet too.”

It is a good instinct and it has the same flaw every hand-kept tracker has: it is only true for as long as somebody keeps typing into it. The engagements kept moving. The spreadsheet did not.

Track every live deal without a pipeline to update

New opportunity, in progress, follow up, won, lost. Real stages, and accurate for about a week after somebody sits down with them.

Clarity reads the mail and moves the deal instead. 43 live deals sit on his board, and not one was typed into a form or dragged between columns. Over 180 days 446 replies went out that he did not have to write.

“I'm a believer.”