Give the admin to software and keep your time for founders. Automate data capture, scheduling and signal monitoring, never ask founders for numbers you already have, and keep crises, key hires and fundraises for a person. Scaling well means you act on signals early and help founders where they already work.
The growth usually comes before the headcount. The fund raises a bigger vehicle, the portfolio goes from 15 companies to 40, and the platform team stays at one or two people. What felt personal at 15 companies starts to feel like a queue at 40.
The instinct is to work harder or to hire. Honestly, I don't think either one fixes it. With good data and the admin handed to software, a team of 3 can do what a team of 30 used to. The real problem is that every task gets the same treatment. A request for a legal template gets the same attention as a founder whose cofounder just quit. Platform time belongs with founders, and admin belongs to software.
Sort every task by frequency and stakes
The simplest tool I've seen platform leads use is a two-by-two. One axis is how often the task happens. The other is what it costs if you get it wrong or do it slowly.
| Low stakes | High stakes | |
|---|---|---|
| High frequency | Automate it | Draft it, a person reviews |
| Low frequency | Answer once, reuse it | A person owns it, every time |
Here's how common platform work lands on that grid:
- Automate: office hours booking, metric requests and reminders, onboarding emails, portfolio news alerts, capturing the asks buried in founder updates.
- Draft, then review: intro requests, LP update drafts, quarterly portfolio reviews, event invitations. Software writes the first version, and a person reads it before it goes out.
- Answer once, reuse: the vendor perks list, legal templates, the hiring playbook. Write the answer once, then send it into the thread where the founder asked.
- A person, every time: layoffs, cofounder conflict, a bridge round, an executive hire, an acquisition offer, anything personal for the founder.
Put your last month of work into this grid. Most teams find the top-left box eats far more of their week than the bottom-right one. That's backwards (and I'd bet yours does too).
Act on signals, not on the calendar
The best platform teams I talk to are proactive and non-intrusive. They act before a founder asks, and they don't create homework for the founder to get there. At 40 companies, the way to do that is what I call a signal-based platform. It works like signal-based sales: signal → reach out → log it.
Some signals worth acting on:
- A founder posts a new role on LinkedIn. Send two candidates that day.
- An ask sits on page 9 of a quarterly update, where nobody who could help will ever see it.
- The same hiring problem shows up three quarters in a row.
- Burn jumps or runway drops below a threshold you set, say 12 months.
And silence doesn't mean nothing's happening. If you haven't heard from a founder in two months, you assume it's quiet. But what if they've been posting about a launch every week?
Keep the list short. Too much signal gets tuned out, and most news alerts are noise. Each signal should create a task for a named person, not just a notification. The machine finds the signal, and a person decides whether it needs a call, a note or nothing.
Meet founders where they already work
The usual advice is to build a resource hub: one Notion page with every template, linked in onboarding. I'd skip it. Platform leads tell me they say "check out our resource hub," and the founder has no idea what they mean. Founders don't log in to portals. They live in email, Slack and WhatsApp.
Keep the material, but deliver it where the question comes up. If a founder asks for an employment agreement in the portfolio Slack, answer in that thread with the file. If three founders asked this quarter, write the answer once and paste it the fourth time. The library is for your team to find things fast, not a destination for founders.
Onboarding works the same way. Skip the welcome packet with ten links (nobody clicks them). Send one email from a real person: who to contact, what you already know about the company, and the one or two things you can help with in the first month. Then do them. Founders notice follow-through more than polish.
Never ask founders for data you already have
I was on the other side of this one. As a startup chief of staff, I wrote our investor updates, and then a stack of forms and spreadsheets showed up asking for the same numbers we'd just sent. I hate forms.
Portfolio data collection scales worst by hand, and it's where firms most often give founders homework. Say your fund has 40 companies and wants six metrics a quarter. The usual plan is a form or an Excel template, then weeks of chasing. But founders already report. The numbers sit in board decks, updates and email threads, sometimes on page 37 of a deck.
A request that respects that has four traits:
- It comes from a real person's inbox, not a no-reply address.
- It's pre-filled with what you already have from decks and updates.
- It asks only for what's missing.
- It accepts a reply in any format: a sentence, a deck, a forwarded update.
It also makes the fund look like it did its homework, and founders reply far more often when they only have to confirm a few numbers.
This is the part of platform work Cura was built for. It sends metric requests from your own inbox, pre-fills what it already has, and asks founders only to confirm or fill gaps. It follows up on its own, and every number links back to its source, so the team spot-checks instead of redoing the work.
Protect the moments that need a person
Write down the moments that always get a direct, human response. Keep the list short and specific, and share it with your partners so everyone agrees on it. Mine would look like this:
- A founder tells you they're running out of runway or planning layoffs.
- A cofounder or senior executive leaves.
- The company starts a fundraise or gets an acquisition offer.
- The company is hiring a C-level or VP role.
- A founder has a personal event: a new baby, an illness, a loss.
For each one, decide who reaches out and how fast. "Within one business day, by phone, from the board partner or the head of platform" is a rule. "We're here if you need us" isn't.
These moments are where founders form their opinion of your firm. Platform leads tell me the same thing over and over: founders forget the newsletter, but they remember who called the week the round fell apart. That's the value-add gap in practice.
A worked example: 30 companies, two people
Say you run platform at a seed fund with 30 active companies and one colleague. A weekly time budget that keeps the human touch could look like this:
| Work | Mode | Hours per week (both people) |
|---|---|---|
| Portfolio news and signal review | Automated digest, 30-minute scan | 1 |
| Metric collection | Automated, review exceptions | 1 |
| Office hours | Self-booked, you attend | 4 |
| Intro requests | Asks captured, you write the intros | 4 |
| High-touch moments | Person, every time | 8 to 12 |
| Programs and events | Planned, partly drafted by software | 8 |
| Value-add log | Intros and hires captured, monthly check | 1 |
The numbers are illustrative. What matters is the shape: the high-touch row should be the biggest single line. If it's the smallest, the repeatable work is still crowding it out. For more on how AI changes this balance, see how AI is changing the role of a VC platform team.
What parts of VC platform work should be automated?
Automate work that is frequent, predictable and low in stakes: metric collection and follow-ups, office hours scheduling, onboarding emails, portfolio news monitoring and capturing founder asks from updates. If a mistake would only cost a few minutes to fix and the task repeats every week or quarter, it is a strong candidate.
Here's my quick test. Write the task out as steps. If the steps are the same every time and don't need judgment, automate it. If one step needs judgment, automate the rest and keep that step for a person. Intro requests are a good example. Spotting the ask and tracking the intro can be automated, but the double-opt-in note should come from someone who knows both sides. So software drafts it, and a person sends it.
Don't automate anything a founder would read as a substitute for attention: condolences, congratulations on a hard-won milestone, or a reply to a founder who's clearly stressed.
How do you keep founder relationships personal at scale?
Keep a short list of moments that always get a person, show up with context instead of generic check-ins, and use signals to tell you when to reach out. Founders judge you on the few times they really needed help, not on how many emails you sent.
Context is what makes a short note feel personal, and founders can tell in one line whether you've been paying attention. "Saw you shipped the new onboarding flow, and your churn number from last quarter suggests it was needed. Want an intro to the head of growth at a company that solved this?" takes two minutes to write if you already know the context. Without it, you're stuck with "How's it going?"
Two habits help:
- Let the record keep itself. Emails, decks and meeting notes should land in one company record without anyone forwarding or logging them, because people who forward by hand forget (I certainly did). Before the next touchpoint, read the record and check what you promised. A promise made in the sixth of nine back-to-back meetings is the one that gets lost.
- Batch your outreach by trigger, not by calendar. A quarterly check-in to all 30 companies feels like a mail merge. A note sent the day after a big hire feels like you're paying attention.