As a Venture Capital (“VC”) investor, making timely and data-driven investment decisions sits at the core of what we do.

Yet, we often come across common problems within our investment process. These have traditionally created bottlenecks, inefficiencies and ultimately sometimes lead us to making untimely or wrong investment decisions.

The traditional VC investment process

Deal origination: The deal sourcing

These initally come from warm intros, investor/founder referrals or cold outreach. Our ideal pipeline scenario would be “high-quality, low-volume” but in reality it is sometimes the other way around. As a boutique venture fund, approximately 100-120 deals would come across our table each month.

Deal screening: The deal vetting

The 3-minute pitch deck scan and/or video review. This process has been labour-intensive, yet cognitively bias. We generally vet down to about 10% of the pipeline – so anywhere between 10-12 projects each month.

Deal evaluation: The due diligence

The investor-founder interaction really begins here – from initial investment meet, Q&A calls to data room audit and length investment memo drafts. Information and communication are often fragmented. After an initial meet, we generally further the evaluation process with up to 3 projects each month.

Also Read: SEA’s venture capital shifts from mega-rounds to AI and SaaS

AI architecture behind a revamped deal screening and evaluation process

How AI has helped as a productivity tool

With AI, the high-impact, quick win for us has really been on the productivity side around screening and evaluation.

We were able to build a proprietary vector database that utilises:

LLM which helps convert unstructured pitch decks and financial models into structured data
Data engineering that extracts, transforms and loads API feed data across a variety of social platforms. It can help identify for instance, a high profile technical talent that has just announced a new stealth project, adding a prospect in additional to our traditional pipeline.
RAG and prompt context engineering to transcribe audio and video data

Also Read: Good ideas are everywhere, venture capital isn’t

The above enables our in-house LLM evaluation matrix to do its work, creating beneficial impact via actions such as:

Benchmarking market sizing and sector analysis in pitch decks to external database
Tracking digital footprint of startups from product development to hiring
Detecting contradictions from a founder’s statement versus a call transcript from two months ago

Next step: The decision making

The investment decision remains an area that needs to be worked and finetuned.

A part of that is on us being able to quantify and accurately input the drivers behind our judgement and decision making.

But as a VC investor, AI has made our investing interesting not only from the exciting projects we see in the sector, but also making our lives easier from a productivity perspective.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post How I use AI in venture capital appeared first on e27.