Inside a PE AI incubator: how Hg Catalyst builds with portfolio companies

Written by
Chuck Ansbacher
Last updated
Share on
table of contents
Down arrow

Tristan Ratchford spent five years at Datadog leading the team behind Bits AI SRE, the company's autonomous AI agent. Now he’s the CTO of Hg Catalyst, the AI incubator at private equity firm Hg, and he was about four months into the role when he sat down for a recent episode of Carried Interest, Affinity's private equity interview series.

This is his account of how a PE AI incubator works.

"Every AI company right now is fighting for distribution, but private equity already has it." — Tristan Ratchford, CTO, Hg Catalyst

What is a PE AI incubator?

A PE AI incubator is a team inside a private equity firm that builds AI products with the firm's portfolio companies and hands them over to run on their own. Hg Catalyst is one example, and Ratchford describes it as one of the first to build this model.

Catalyst starts from a thought experiment about the portfolio company itself: What if this were an AI-native startup?

The question shifts the work from adding an AI feature to asking what the business would look like if it were designed around AI from the start.

How a PE firm helps portfolio companies build AI products

A Catalyst engagement runs on a small team, and it begins with a workshop before any building starts. According to Ratchford, this is typically two engineers from the Catalyst team and a PM, with Ratchford himself overseeing it.

Next, a temporary team forms and portfolio company staff are pulled off business as usual to work on the build.

Here’s what an engagement, in the order Ratchford describes it:

  1. A workshop with the portfolio company.
  2. A temporary team of two Catalyst engineers and a product manager, overseen by Ratchford.
  3. Portfolio company staff pulled off business as usual to join the team.

AI incubator vs. consulting and forward deployed engineering

An AI incubator builds the product and stays responsible for it after launch, which separates it from consulting. "It's not like consulting, where you prove value and bounce," Ratchford says, drawing the line directly. “We have a responsibility to make sure it runs afterwards.”

Forward deployed engineering, on the other hand, is where an engineer embeds with a customer for the long term. Catalyst's version is essentially a middle ground between the two. A temporary team that builds, hands over, and then the customer is paired with two enablement programs—Academy and Jumpstart—before the team leaves. Ratchford describes the moment of transfer like a baton is being handed over.

This approach works because a private equity portfolio doesn’t look like one company. Ratchford describes the range Catalyst works across as “60 different environments, 60 different tech stacks, 60 different problems."

A team that builds and hands over has to work across each of those environments.

Where could a PE AI incubator go next?

Ratchford's hope is that the incubator becomes part of how Hg competes for deals. "Medium term, I would love to see Catalyst become a differentiator in deals,” he explains.

In Ratchford's words, distribution is what every AI company is fighting for, and private equity already has it through its portfolio. Catalyst is built to use that advantage.

Enjoyed this? More from the Carried Interest Series

Related Articles

Inside a PE AI incubator: how Hg Catalyst builds with portfolio companies

Product release recap: October

What shipped in October: set your firm's rules once

The deal has a deadline. The relationship doesn't.

The deal has a deadline. The relationship doesn't.

decorative