Happy Thursday and welcome to another edition of Beyond TVPI.

Today, we’ll go into the way we see ourselves as LPs. Fiduciaries, salespeople, or both? I’ll argue the latter.

After that, I’ll cover why new AI model releases don’t excite me as much anymore. Not because AI matters less. But because I think the real breakthrough has shifted elsewhere.

Before that, a quick note: last week we published a skill for LPs (and, as it turns out, GPs) to map the market.

The feedback has been strong - including use cases we hadn’t explicitly tested. One example: “Industrial US Managers,” which came back with a “pretty darn good list”.

If you are using Claude, ChatGPT or CoPilot, you can get it here.

BV Investment Partners just raised $2.5 billion in a one and done close. Parthenon, Nautic, Percheron - the same stories. If you weren’t already in those rooms, you weren’t getting in this cycle.

That’s not a complaint. That’s the asymmetry.

The best GPs don’t need you. You need them.

By the time the fundraise starts, the book is already decided.

Most LPs flinch at being called salespeople. We see ourselves as analysts, gatekeepers, allocators of capital. That self-image is precisely the problem.

The best GPs are closing books with LPs who started selling them on the relationship two years ago.

It took me a long time to get comfortable with that framing myself. The word sales sits awkwardly next to fiduciary, and it took years of watching how the market operates before I accepted that my discomfort with the framing was a heavy cost to bear.

I’ve written before about the Alpine miss - passing in 2018 because the unrealized marks made us uncomfortable. The piece focused on what we should have done after the pass.

But there’s another interpretation.

We didn’t pass because we misunderstood the data.
We passed because we started too late to trust it.

Had we been tracking Alpine for two or three years before they came to market, those concerns might already have been resolved. We would have watched the marks convert. Heard the story compound.

That Alpine fund is going to be one of the best 2019 vintage funds.

That’s the price tag on cold-meeting a manager who would have made sense with a few years of context.

Imagine a GP told you, with a straight face: “We wait for the banker to call us, then we evaluate.”

You’d cross them off the list immediately.

The mirror is exact.

In every intermediated market - PE deal flow, real estate, executive recruiting - the participants who win are the ones building relationships before the auction. Capital is just another commodity. The only edge is showing up early and consistently.

So ask yourself: Why shouldn’t this logic apply to the ones allocation the capital as well?

Sidebar: This was the process for getting into a manager most readers will know. It sounds simple. It is. But it is not easy.

We stopped by Redacted’s offices every 6-9 months. Not because we had to. Not because they’d invited us. Because that was the cadence we’d assigned them - and we stuck to it. I often brought new colleagues to ensure we asked all the right questions - and still rated them A.

In between visits, we treated them like an existing portfolio fund. Track record dissection. Sector exposure. Team movements. Returns by vintage. Not for an IC - for an ongoing A (and frankly A+) manager.

By the time of one meeting, maybe a year before the fundraise, it was just me and the MD. We’d scheduled an hour. It took 35 minutes.

Not much had changed. The fund was humming, the team was stable, the cases were tracking. We could have stretched the conversation. Instead, I chose to respect his time.

When the fundraise finally opened, we still ran our process - calls, references, framework. But the decision was effectively made.

Not because we skipped the work. Because three years of unremarkable persistence had already done most of it.

The LPs who got in had been showing up for years. The LPs who started showing up when the fundraise officially opened never got a meeting.

Without a system, “stay close” becomes a sentiment, not a behavior.

FundFrame is built to make that behavior systematic.

Learn more here.

If you accept this, a few questions:

Most LPs, of course, do some of this - but most of us can do more. And that’s great - that’s your opportunity!

So take the manager you most want access to. Ask if they’re at SuperReturn.

If nothing else, it will likely turn into a Zoom before the summer.

And it’s probably a better use of your time this week than another market map.

New AI model releases don’t really get me excited anymore.

Skills do.

We use AI constantly at FundFrame. But over time I realized something:

A prompt gives you one output.
The conversation ends.
The lesson dies with it.

Next time, you start from scratch again.

That doesn’t scale.

What scales is a skill.

A written-down procedure that teaches an AI agent how to do something the way we do it.

How to read between the lines in a GP quarterly report.
How to interpret waterfall language in an LPA.
How to conduct a market map search.

Each one captures judgment that used to live in someone’s head and turns it into something callable, improvable, and durable.

We’re not really writing prompts anymore.

We’re writing intelligent workflow software - in English.

That’s the most excited I’ve been about AI in years.

And now we’re starting to build these skills directly into FundFrame.

Three years ago I would have called that a moonshot.

Last week, we started shipping it.

This newsletter is written by Steffen Risager, the founder of FundFrame, a platform for LPs to manage their private markets investments.

Before that, Steffen was CIO at Advantage Investment Partners, a Danish Fund-of-Funds.

Steffen has a decade of experience as an LP, and has made commitments totalling approx. $6bn across fund- and co-investments.