Embed with a clear scope, an end date and a named owner at the company who takes over. Do the work inside the founder's tools and rituals, and build the system the team will keep running. Don't take over decisions, don't become the company's permanent hire, and don't report back to the partners without the founder knowing.
I have a soft spot for embeds. When I was a startup chief of staff, the investors who actually helped us were the ones who spent real time with us and had live context on what was going on. Everyone else gave advice that was a quarter out of date. An embed is the most direct way a platform person can get that context.
It's also the easiest way to damage a relationship. You're the fund's employee sitting inside a company the fund owns part of. The founder wonders what you're reporting back. The team wonders if you're auditing them. And your partners wonder why you've disappeared for six weeks. Most of the rules below exist to handle those three worries.
When an embed is the right call
An embed works when there's a specific job, a deadline and a gap the company can't fill fast enough on its own. It fails when it's a vague offer of help or a quiet rescue mission.
| Good reasons to embed | Bad reasons to embed |
|---|---|
| Run the first sales or engineering hiring sprint | "The founder seems overwhelmed, go help" |
| Build the data room and metrics pack before a raise | Partners want a closer look at a company they're worried about |
| Set up board reporting and a monthly close with the finance lead | Covering a leadership gap with no plan to hire |
| Stand up a pricing or packaging project with the founder | The platform person wants operating experience |
| Hold a function together during a short exec search | Nobody else at the fund has time |
The second column isn't always wrong forever. But each one needs a different fix first: a partner conversation, a search, or a coach.
Write a one-page embed agreement
Before day one, write a single page and agree it with the founder. Keep it plain, and share it with the partner on the deal too.
- Goal: the one outcome that marks success. "Two account executives hired and onboarded."
- Deliverables: what exists when you leave. A hiring process, a scorecard, an interview loop, a runbook.
- Dates: start, halfway check-in and end.
- Company owner: the person who takes over the work when you leave.
- Time: how many days a week you're there, and when you're in person.
- Access: which tools, channels and documents you'll use.
- Information line: what you'll share with the fund, and how the founder sees it first.
- Out of scope: what you won't do. Firing, comp decisions and board matters usually belong here.
That last line saves you from scope creep in week three, when the founder asks you to "just sit in" on a tough conversation.
The do's
Work inside their tools. Use the company's Slack, their ATS, their docs. Don't set up a parallel Notion page that dies when you leave. The system has to live where the team already works.
Show up in person early. The first week sets the tone. Sit with the team, go to their standups, and learn the rhythm before you change anything.
Start by doing, then teach. Run the first few hiring screens yourself, then have the owner run them with you watching. People learn a process faster by seeing it work once.
Write things down as you go. Every template, script and checklist you create goes into a shared folder the company owns. If you get pulled away tomorrow, the work should survive.
Keep the founder close. A 20-minute check-in every week, even when nothing's wrong. Bring what you've done, what's stuck and what you need from them.
Keep your other founders covered. Tell the rest of the portfolio you're on an embed, and set up who handles their requests while you're out. Embeds go wrong when the platform person vanishes for everyone else.
The don'ts
Don't make the founder's decisions. You can recommend a candidate. You can't hire them. Once the team starts asking you instead of the founder, step back.
Don't become the hire. If the company starts treating you as its head of talent, the embed has turned into free headcount. That's bad for them (they're not building the team they need) and bad for you.
Don't audit the team. You'll see things that aren't working. Tell the founder, privately. Don't build a case file.
Don't freelance on fund business. If someone at the company asks about the next round or the cap table, send them to the founder or the partner.
Don't let it run past the end date by default. An embed with no end date turns into a dependency.
Handle the information line carefully
This is where embeds fail most often. Partners will ask you, "How's it really going over there?" And you'll know more than anyone else at the fund.
Set one rule and say it to both sides on day one: nothing goes to the partners that the founder hasn't seen first. If you have a concern about a leader, raise it with the founder, then tell the founder you'll mention it to the partner. Most founders accept that as long as there are no surprises.
A partner update can be short and shared with the founder in the same thread:
Week 4 of 8 at the company. Hiring sprint: 40 candidates screened, 6 in final rounds, 1 offer out. The sales lead now runs screens alone. Open issue: the comp band for senior AEs is below market, and the founder is deciding this week. No help needed from you yet.
That message is illustrative, but the shape works. It covers progress, the handover and one honest issue, with no gossip.
How long should a platform person embed with a portfolio company?
Aim for four to twelve weeks, with a check-in at the halfway point. Shorter than that and you can't learn the context. Longer than that and the company starts to depend on you. Set the end date on day one, and extend only with a new scope.
The right length depends on the job. A data room before a raise might take four weeks. A first hiring sprint for several roles is closer to eight to twelve, because candidates move at their own pace. Use the halfway check-in to decide: on track, cut scope, or end early. Ending early is fine if the owner is ready. It's a good sign, not a failure.
Part-time embeds work too. Two days a week for eight weeks often beats one full-time month, because the team has time to try things between your visits.
How do you leave an embed without leaving a gap?
Plan the exit from the first week. Name the person at the company who will own the work, have them shadow you, then switch roles so you shadow them. Leave a short written runbook and a 30-day check-in on the calendar. If nobody can own it, the company needs a hire, not more of you.
A simple handover sequence:
- Weeks 1 to 2: you run it, the owner watches.
- Middle weeks: the owner runs it, you watch and fix.
- Last week: the owner runs it alone. You're on call.
- Day 30 after you leave: a 30-minute check-in. What broke, and what did they change?
The runbook should fit on two pages: what happens each week, who owns each step, where the templates live, and who to call when something breaks. If you find yourself writing ten pages, the process is too complicated for the team to keep running.
After the embed, write a short note for your own team on what worked. Over time, those notes become the playbook for the next embed, and they're good evidence of value when partners ask what the platform team did this year. If you're newer to the role and weighing whether to take on an embed at all, the first 90 days in a VC platform role covers how to pick your early projects.