We stand at the brink of the next technological revolution, where the rapid adoption of AI is reshaping industries at an unprecedented pace. From healthcare to construction, and even in complex, service-based sectors like venture capital, AI's impact is profound. The question is no longer if AI will transform these industries, but how quickly and effectively. Yet, I regularly find myself in conversations where people recognize its potential but struggle with where to begin.
After spending the last two years meeting entrepreneurs building vertical-specific solutions with outstanding outcomes, we were eager to adopt this technology to automate some of our own day-to-day workflows. Since launching our proprietary AI-driven platform, we've achieved remarkable results:
- 10X increase in top-of-funnel opportunities: This expansion enabled us to maximize market coverage across all our investment theses.
- 5X boost in team efficiency: Our team can now review potential opportunities much faster, focusing on the most promising ones.
Where to Begin
Every organization's automation journey is unique, and determining which part of the value chain to focus on can be daunting. For venture capital firms, this often means choosing between sourcing, screening, diligence, or portfolio/relationship management. Here are some key factors to consider when deciding where to begin:
- Investment strategy: Are you a generalist or thesis-driven?
- Portfolio size: Funds with large portfolios may prefer to automate KPI tracking.
- Team size: Smaller teams may lack the analyst resources and therefore want more help automating sourcing.
- Stage focus: Early-stage investors cannot solely rely on Crunchbase or Pitchbook for sourcing and may consider additional data sources.
Identify which organizational challenges you hope to solve, such as growing top-of-funnel, faster deal review, automating founder outreach, diligence support, or managing an existing portfolio.
At Ardent, we zeroed in on sourcing and screening, particularly at the top-of-funnel, to drive maximum value. With our thesis-driven early-stage strategy, we aimed to cover every deal in our investment areas and efficiently manage the surge in opportunities. This ensures our small but mighty team focuses on the right opportunities within each thesis.
Moving Beyond What's Available Off the Shelf
Once you've identified your starting point, it's time to enhance your existing tools. While we leverage Affinity, Harmonic, and others in our "VC stack" (and are constantly reviewing which to keep/drop/add), we built bespoke software to achieve better outcomes.
One of our ML models, which predicts whether a company is an Ardent thesis fit, has achieved ~95% accuracy, significantly boosting our efficiency. Prior to this model, manual reviews were necessary to determine thesis fit.
Our v2 platform (still in testing) prioritizes opportunities identified as "thesis fits." We created agent-based workflows to mimic our team's existing review process — analyzing company websites, founder profiles, and third-party data to determine if they met various criteria to be considered interesting and highly relevant to our investment team. It turns out generative AI is excellent at automating the aggregation and summary of this information and has made the human review of the funnel even faster and more effective than v1.
Here is how our sourcing and screening have evolved since launching our internal tools:
Adapting Processes to Fit Our New Tooling
Tools alone aren't enough; processes must evolve as well. When we realized our platform had such a profound impact on how we operated, we quickly realized that we needed to rethink the role of our analysts, how we ran our pipeline meetings, how we assigned opportunities to deal teams, how we used our CRM, etc.
Some processes became automated and disappeared, while others needed new workflows to prevent team overload and maintain our advantages. For example, we created human-in-the-loop feedback workflows to ensure high model accuracy:
- If an opportunity was incorrectly identified as a thesis fit, we manually flag it so our model can learn. We also look at the inverse cases, which means we need to take samples of all the opportunities the model has reviewed and confirm predictions.
- We also now capture how far opportunities make it in our funnel, "pass reasons," and additional notes about companies, founders, markets, etc., that we feel are relevant and could help the model better understand the rationale for what we think is interesting about a company or a space.
The role of an Ardent analyst is now a little different than those at other funds. Our team can simply log into our platform and immediately spend more time reviewing a backlog of highly curated companies to review and reach out to instead of culling through the noise if they were to just start searching on the company/sourcing databases and manually review each website to determine whether something was interesting.
Future outlook of venture capital and automation
I'm optimistic that AI won't take all of our jobs, but I do believe that funds of the future that leverage automation will look and operate differently than those of today. It's only a matter of time before we see more funds with in-house technologists and fewer junior analysts.
Not every fund's journey or approach will look similar to Ardent's. I wanted to share what is working for us with the community to spark conversation and provide some (hopefully) helpful insights for those looking to begin their journey. Interested in sharing notes? Reach out! dp@ardent.vc
