Capture every touchpoint, including events, office hours, intros and founder requests, from the email, calendar and chat where it already happens, and centralize it in one searchable system tied to portfolio companies. Over time this reveals portfolio patterns, common needs and expert networks, making the platform function more valuable to founders and partners every year.
Most platform teams I talk to already know a huge amount. Which founders struggle with enterprise pricing. Which advisor gives great feedback on sales comp. Which intros turned into customers. The problem is where that knowledge lives: in one person's head, their WhatsApp, a few DMs and a spreadsheet nobody's opened since March.
That works until a partner asks "which of our companies need help with US hiring?" and you spend a day finding out. And it falls apart when someone on the team leaves, because the knowledge walks out with them.
Don't ask anyone to log anything
The usual fix is a logging habit: a form after every office hours session, a Slack shortcut for every request, a rule that everyone updates the CRM within a day. I've watched a lot of teams try this. It holds for a few weeks, then it decays. People who have to forward or log things by hand forget, and the gaps land exactly where the week was busiest.
The better design is ambient capture. Almost every platform interaction already leaves a trail in email, calendar, Slack or a group chat. Pull from that trail instead of asking people to recreate it.
- Founder requests. Most arrive by email or Slack, and many hide in monthly updates as one line near the bottom. Capture them from the source, including the asks nobody sent to you directly.
- Intros. The double opt-in email is the record, and the reply a few weeks later is the outcome.
- Events. The calendar invite has the guest list. The follow-up emails you send the next morning hold the promises people made.
- Office hours. The invite has the company and the expert. The expert's follow-up note has the takeaway.
- Partner meetings. Investors forget what they promised in the first of nine back-to-back meetings. If the meeting notes land in the same system, the promise turns into a task instead of a lost favor.
Be honest about the hard part. WhatsApp threads, texts and Slack DMs (what I call last-mile data) are the hardest to capture, and they're often where the best context lives. Start with email and calendar, which cover most of it, and add the rest as you can. A partial context layer still beats none. Just know where its blind spots are.
Define a small set of record types
Capture is only useful if it lands in a shape you can query. Pick five record types that cover most of your work, and give each one a handful of fields.
| Record type | Fields |
|---|---|
| Founder request | Company, date, topic tag, what they asked, status, who resolved it |
| Intro | Company, person introduced, reason, date, outcome after 30 days |
| Event attendance | Event, company, attendee role, follow-ups promised |
| Office hours | Company, expert, topic tag, one-line takeaway |
| Expert contribution | Expert name, area, companies helped, founder rating |
Two rules make the data useful later. Every record links to a portfolio company, so you can search by company. And every record links to its source, the email or message it came from, so anyone can check it in one click instead of asking around. Notion, Airtable or a CRM like Affinity or Attio can all hold this. Whatever you pick, make sure the data stays yours and you can export it to your own warehouse.
When you do need to ask a founder something, like an intro outcome, keep it light:
"Quick check on the intro to the head of procurement from a few weeks back. Did you two connect, and is anything happening there? A one-word answer is perfect."
Founders answer these when they're short and specific. But only ask for what you can't already see. If the answer already sits in last month's update or a board deck, pull it from there. You shouldn't ask for things a founder has already given you.
Use a fixed tag list
Free-text notes are hard to search. A fixed list of topic tags makes patterns visible. Start with 10 to 15 tags, for example:
- Hiring and recruiting
- Sales and go-to-market
- Pricing
- Fundraising
- Legal and compliance
- Finance and accounting
- Product and engineering
- International expansion
- Marketing and brand
- People and culture
Review the tags once a month. If "other" is used often, split it into a new tag. If two tags always appear together, merge them. If a tool applies tags for you, spot-check a sample each month. Don't trust it blind.
Turn the data into answers and a value-add log
Data turns into intelligence when someone asks it questions on a schedule. Once a quarter, pull the records and answer these:
- What were the top five request topics, and which companies asked?
- Which experts got used most, and how did founders rate them?
- How many intros led to an outcome within 30 days?
- Which companies had no platform touchpoint in the last 90 days?
- Which events produced follow-ups that actually happened?
Here's a worked example. Say your fund has 30 portfolio companies and a two-person platform team. In the quarterly review you notice that eight companies asked about usage-based pricing in three months. That's a pattern, not eight separate requests. Instead of eight one-off calls, you run one pricing workshop with your best expert and record it. The next company that asks gets the recording in a reply and an offer to talk.
Question four is the one people skip. A quiet founder isn't always a founder who's fine. Sometimes they stopped asking because help didn't arrive last time.
The same records give you a value-add log: a running count of the customer, talent and investor intros the platform made, and what came of them. Partners use it in competitive deals and pro-rata conversations, and it answers the LP diligence question about how the firm actually helps. The outcome column is what turns a list of intros into proof, which is the evidence partners need in competitive deals. For events, this is also the step that separates a nice dinner from one with lasting impact, as covered in how VC firms design events with lasting impact.
Run a signal-based platform
Your own touchpoints are only half of it. A request log tells you what founders ask for. It doesn't tell you that a company just posted three sales roles, lost its CTO or launched in a new market. Sales teams have a name for acting on this kind of trigger: signal-based selling. Platform teams can run the same loop: spot the signal → find something specific to offer → reach out before the founder asks.
Be choosy. Most news alerts are noise, and a team flooded with signal starts ignoring all of it. The useful signals are the ones that point to a concrete offer: a role you can send candidates for, a market where you know a customer, a hiring struggle that shows up in three monthly updates in a row.
This is the part Cura is built for. It pulls portfolio company signal from LinkedIn, X, news, board decks and email into one feed and creates follow-up tasks tied to each company. Its MCP server lets an AI assistant like Claude or ChatGPT answer questions across that data. For the broader idea, see what portfolio intelligence is.
Settle trust before tools
In my experience, the blocker here is rarely technical. It's psychological, and it comes down to trust. A shared system only works if partners let their email and calendar feed it, and if the team agrees that platform knowledge belongs to the firm, not to whoever holds the relationship. Get that agreement first.
Founders tell the platform team things they may not tell their board. Treat that with care.
- Decide who can see which records. Sensitive requests, like a co-founder conflict, should not be visible to the whole firm.
- Don't keep personal details beyond what you need to help.
- Tell founders what you track and why. Most founders welcome it once they see it means faster, better help.
What data should VC platform teams collect?
Collect who asked for what, what you did, who helped and what happened next. That covers founder requests, intros and their outcomes, event attendance and follow-ups, office hours topics and expert contributions. Tie each record to a company and to the email or message it came from, so patterns show up and anyone can check an entry against its source.
If you only track one thing, track outcomes. A count of intros made says how busy you were. A count of intros that led to a customer, a hire or an investor meeting says what the platform is worth. For a broader approach to keeping tabs on each company, see how to track portfolio companies.
How do you build a portfolio knowledge base?
Start with one shared database tied to portfolio companies and a fixed set of tags. Feed it automatically from the email, calendar and chat where the work already happens, not from manual logs. Review tags monthly, and turn the most common answers into material you send the next founder who asks.
A crawl, walk, run rollout keeps it manageable:
- Crawl (month one). Connect email and calendar so founder requests and intros land on their own. One database, linked to companies.
- Walk (months two and three). Add events, office hours and partner meeting notes. Start the monthly tag review.
- Run (quarter two onward). Run the quarterly review, send a short value-add summary to partners, and start acting on outside signal.
After a year, you'll have something no new hire and no competitor can copy quickly: a record of what your portfolio needs and who in your network can help.