What Venture Capital Firms Need Before Using AI

How quickly could your team pull together a standard portfolio reporting metric?

Instantly? In an hour? Or maybe longer? At Venture Intelligence Day in Prague, when this question was asked in a room filled with professionals from finance, data, and investment industries, the result was quite astonishing: Only a handful could access a standard metric within minutes.

For Yun Yun Lin, Senior Data Expert at Trinsik, this highlighted a challenge that is often overlooked in conversions around AI, which is: 

Before organizations can generate better insights, they need to be able to access and trust the data they already have. 

As Yun observed:

A quick show of hands said it all: even standard metrics still take real effort to produce, and the same KPI often shows different numbers in different reports. What the industry needs is one central source of truth, where both the data and its definitions live in a structured way.

Not every VC firm is starting from the same place.

AI is increasingly becoming an important topic in venture capital - for portfolio monitoring, reporting, and investment decisions. The discussions showed that VC firms are currently at two different stages when it comes to their data infrastructure. Some already have a data warehouse in place. For these firms, the next step is adding more context to their existing data. This includes more qualitative information, knowledge, and insights to interpret numbers and generate more meaningful conclusions with precision.

The issue is that most firms are still missing that foundation. The priority for them should not be introducing new AI tools, as this may not give accurate results. The best order of action is building a data warehouse, bringing their data together, organizing it well, and ensuring consistency. As Yun puts it: 

Last year, a data warehouse was a reporting tool for large VCs. This year, it is the foundation for anyone using AI.

What getting the right foundation actually looks like:

During the workshop, Trinsik presented its work with the EIC (European Innovation Council) to show why a well-structured data warehouse is essential and what it can empower.

The concept is relatively straightforward: bring information from different sources together, organize it consistently, and make it easier to access and use. In practice, this means:

• Centralized data: Working with information spread across different systems and spaces can be a pain. All data is collected and structured centrally.

• Language consistency: Definitions are agreed upon and documented, which makes for more reliable reporting.

• Ability to reproduce reports: Versioned snapshots allow organizations to recreate previous reports and understand what the data looked like at a specific point in time.

• Less manual work: Repetitive tasks are replaced by automated data flows and quality checks, which means fewer manual number checks and copying information from one place to another.

Although these process changes seem like mere technical improvements, they make big differences in how organizations work with information. Yun shares a valuable example from the EIC:

At the EIC, the team used to spend weeks copying data from one sheet to another. Today that time goes into asking better questions. Structured data lets automation and agents do the heavy lifting, so people can focus on the insights only they can bring.

Better AI starts with better data

AI has the potential to change how VCs understand their data, monitor investments, and make better, more informed decisions. However, it cannot make up for inconsistent, unreliable data scattered across different systems.

For some firms, the next step is enriching their existing data with more context. For others, it means going back to the basics and building a reliable data foundation first.

The biggest takeaway from Venture Intelligence Day? Before expecting AI to give superior answers, organizations need to make sure they're working with better data to begin with.