Customer story

Why New Zealand’s most prominent AI engineer decided to buy vs. build his portfolio management

With Jin Budelmann, Head of Data + AI

CuraGD1
portfolio companies on Cura
~42
funds in one place
4
tools GD1 built on Cura
3
Firm
GD1, generalist VC, New Zealand
Portfolio
~42 companies across four funds
Switched from
Fundwave and Excel workbooks
On Cura since
October 2026

Jin Budelmann could have built this.

Jin spent 15 years as a software engineer across New Zealand startups and corporates. Jin runs the Prompt Poets Society, the country’s largest independent AI meetup. At GD1, a generalist venture firm in New Zealand, Jin’s mandate is to use AI to make the whole firm faster.

A system for portfolio data looks like the first thing an engineer like that would build. Jin bought it. GD1 now runs its portfolio data on Cura, and Jin builds GD1’s own tools on top.

“The cost is hidden for a lot of people because the upfront cost is so cheap. You can get to good results, but those systems don’t perform consistently and they don’t scale. Portfolio monitoring wasn’t something I wanted to build or maintain for those reasons.”
Jin Budelmann, Head of Data + AI, GD1

From Fundwave and Excel to one source

GD1’s portfolio data lived in Fundwave and a set of Excel workbooks. Neither told the full story.

Founders were asked to report into Fundwave. Few did, and those who replied used whatever format they liked. The team typed each reply into a workbook, worked out the details, and updated the quarterly reporting by hand.

“Fundwave never gets fully updated, so it’s only got partial data. Nobody looks at it.”
Jin Budelmann, Head of Data + AI, GD1

Before Cura

  • Founders asked to report into Fundwave. Few did.
  • Each reply typed again into an Excel workbook.
  • Quarterly LP reporting updated by hand.
  • No LP newsletter. No one had time to gather the content.

Partial data that nobody looked at.

With Cura

  • Cura sends GD1’s information requests to founders.
  • Cura keeps every metric in one place.
  • Quarterly LP reporting runs on Cura’s metrics.
  • Claude drafts the newsletter from Cura’s portfolio news.

One source that Jin builds on.

Build what is unique. Buy what is hard.

Jin’s rule has two parts. Build the parts that are unique to how GD1 works. Buy the parts that are hard to keep right.

The data layer is the hard part. A one-off automation is cheap to build and looks good on day one. To keep it accurate, a team must write evals, run them constantly and catch model drift. So GD1 buys three things from Cura: the information requests to founders, the portfolio metrics, and public news on every company.

“I don’t want to buy a massive software package and mold our processes to fit it. I want to buy the things that really add value, integrate them into a wider system, and customize it to fit exactly how we operate.”
Jin Budelmann, Head of Data + AI, GD1

GD1 looked at other tools before it chose Cura. Onboarding took one step: GD1 added its list of portfolio companies and watched the data flow in.

Quarterly LP reporting on live metrics

Each quarter, Cura sends GD1’s information requests to the founders. The replies land in Cura as portfolio metrics. GD1’s quarterly reporting to LPs now runs on those metrics.

The hand-built workbooks are gone. The investment team also opens Cura directly to check in on their companies.

An LP newsletter, back from the dead

GD1 had stopped sending an LP newsletter. No one had time to gather the content. Of the new one, Jin said: “We couldn’t do that without Cura.”

Jin set up Claude to pull public news on the portfolio from Cura. Claude picks a theme and surfaces a few strong portfolio stories from the past two months. The newsletter goes to existing and prospective LPs, many of them overseas. It shows them the depth of New Zealand’s deep tech ecosystem.

“Nobody has time to write and curate all of that content… Cura gives us a means of automating the portfolio news part of the LP newsletter.”
Jin Budelmann, Head of Data + AI, GD1

A daily portfolio brief in Slack

Every day, a message lands in GD1’s portfolio companies channel. It pulls the last day of public news from Cura, drops anything already reported, and lays out what is new.

Jin called it one of the quickest and most valuable things built so far.

Key takeaways

  • Price the upkeep along with the build. A first version of an AI system is cheap. The cost is in the work to keep it accurate: evals, constant runs and model drift.
  • Buy the data layer. Build the workflows. Buy the parts that are hard to keep right. Spend your engineering time on the parts that make your firm different.
  • Make the software fit your process. Pick tools that you can build on and connect to a wider system. Do not change how your firm works to fit a package.
  • Start with one small tool. GD1’s daily Slack brief was one of the quickest things Jin built, and one of the most valuable.

Thank you to Jin for sharing GD1’s story.

Read more customer stories

For teams that build their own tools

See Cura running on your own portfolio.

GD1 started with a list of portfolio companies. Start with yours. Book a demo and we’ll load your portfolio into Cura, ready for you to build on.

Book a demo