The M&A infrastructure Christine McDannell built for herself

Christine McDannell has built most of a sell-side M&A operating system by herself, and she thinks it should exist as a product. She runs The Magnolia Firm, a brokerage in Austin that sells profitable tech companies, and has cut payroll by 83% while assembling it. Asked whether the infrastructure should be packaged for other firms, her answer was immediate.

"In M&A specifically, this whole infrastructure — that would be rad. That would be rad, because it is very tailored, like what we've built out. It wouldn't need to be changed and adjusted if [it were for] another industry, which wouldn't be hard."

She is one person with a colleague in Toronto. The thing she is describing is a vertical software company.

What has she actually built?

A set of skills on a general-purpose model, running locally for confidentiality. McDannell moved to Claude at the beginning of March after three years on ChatGPT, and runs two accounts at roughly $100 a month each, on a separate Mac mini bought specifically for the firm because the work involves company financials.

What sits on top of it, by her account: an email agent that checks every 30 minutes and drafts; a morning agent that reads her calendar and produces a research brief on each prospective seller; a valuation skill encoding her industry knowledge and her own past deals; due diligence sweeps across public sources; and video editing.

Her framing of the price is the part worth quoting to anyone modelling willingness to pay in this market. She prices what the tool does for her at roughly $40,000 a month of labour, against a $100 subscription.

Why does she think it would generalize?

Because the sell-side sequence is broadly the same at every boutique firm. Galina Fendikevich, who hosts Artificial Reality, the AI buyer-evidence podcast, mapped the standard steps back to her during the interview: preparing the data room, producing the first valuation, running the seller interview, doing buyer outreach. Only the business logic and the underlying data change from firm to firm.

McDannell agreed with that map and added the step she thinks is underrated.

"Even the due diligence. Like I said, I do deeper due diligence than the buyers do, up front, before I take a client ... I'm looking at the Instagram comments, make sure there's not complaints. I'm combing through everything."

The AI version of that is a skill she describes plainly:

"A skill to then tell it to scrape every single thing on this owner and every single thing on this business. I want no surprises — lawsuits, comments, et cetera. Anything bad I need to know up front so that we can figure out the object[ion] — you know, people are gonna find it otherwise."

What would an M&A platform have to cover?

Six things, drawn from what McDannell described doing rather than from any vendor's category definition:

• The standard sell-side sequence — valuation, deep pre-listing due diligence, packaging and narrative, data room, buyer outreach, buyer qualification.
• Encoded industry knowledge and market conditions, rather than requiring every firm to supply its own.
• Adverse-information discovery across public sources, including social comments and litigation.
• Configurability, because her valuation skill draws on her own closed deals and a generic one would not.
• Confidentiality handling — her firm strips company names off financials before submitting them, and runs the work on hardware it owns.
• No domain-specific model. She is explicitly not asking for one, on the reasoning that general models are improving too quickly to justify building it.

Is anyone going to buy it?

That is the uncomfortable part, and it is analysis rather than her testimony. There are a lot of Christine McDannells across professional services right now, each independently rebuilding the same workflow, each unaware the others exist. That looks like a market until you notice the mechanism producing it.

She built her version in months, alone, without writing software, using skills on a subscription product. If every capable boutique firm can now do the same, the addressable market for the packaged platform is the firms that cannot — which is a smaller and less profitable customer than the one a deck usually assumes.

The window is not obviously open for long. Whoever ships this is competing less with an incumbent than with the general-purpose tool their buyer already pays $100 a month for.

What to do with this

If you are building for M&A or anything adjacent, do not start with the data room. Start with pre-listing diligence, which is the step McDannell says she does more deeply than the buyers do and the one she has already automated most aggressively. It has a checkable output, an obvious failure mode, and a buyer who can describe exactly what she wants found.

Then answer the question her own build raises: what does your product do that a competent operator cannot assemble herself in a quarter?

Where this comes from

S1E5: How she cut her payroll by 83% using one AI agent — the full interview with Christine McDannell. Listen or watch: YouTube, Spotify or Apple Podcasts.