Artificial Intelligence in Venture Capital
Enhancement of the venture capital investment process through hybrid human-AI models for pitch-deck evaluation
J.N. de Klerk (TU Delft - Applied Sciences)
Z. Roosenboom-Kwee – Mentor (TU Delft - Technology, Policy and Management)
J. Gartner – Graduation committee member (TU Delft - Technology, Policy and Management)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
The pitch-deck evaluation is a crucial first step in the Venture Capital (VC) investment process, where investors assess a startup’s potential in the first screening of the startup. Traditionally, this relies solely on manual human judgement, making the process time-consuming, subjective and prone to bias. With the rise of Artificial Intelligence (AI), new opportunities emerge to support the decision-making with data-driven insights. However, due to the qualitative nature of early-stage startup evaluations, implementing AI in this context remains challenging due to its quantitative and data-driven nature.
This research therefore explores the potential of hybrid human-AI models (combining human judgement with AI support) in early-stage VC pitch-deck evaluations. A case study was conducted at InnovationQuarter using the AI tool 'Deckmatch'. Through a combination of interviews, a pilot experiment and a follow-up workshop, this study investigates how hybrid models can impact the effectiveness and efficiency of pitch-deck evaluations. This aims to answer the following research question:
How can hybrid models, combining artificial intelligence and human judgement, improve the effectiveness and efficiency of pitch-deck evaluations in early-stage venture capital?
Throughout the research, data was obtained with guidance from the Technology Acceptance Model (TAM) and Behavioral Decision Theory (BDT). The interview dataset consists of six interviews with VC professionals and their perceptions prior-usage of hybrid models within pitch-deck analysis. This was followed by a pilot experiment and workshop including both seven participants. During the pilot experiment the efficiency and efficacy was measured, in addition to the user experience. Afterwards, perceptions on hybrid models after experience with hybrid models was gathered.
It was expected that the hybrid models would enhance both the efficiency and efficacy of the evaluations, leading to a positive attitude towards use. The actual findings suggest that hybrid models show potential to increase the effectiveness or efficiency of pitch-deck analysis depending on the hybrid model used. Here, the sequential and interactive search models show the potential for improved effectiveness, at a slight decrease of time efficiency. On the contrary, the autonomous search model shows the potential for an improved time efficiency at a lower effectiveness. For the usage and preference of hybrid model type, the trust and transparency in the AI output is highlighted to be of importance. These findings contribute to the real-world understanding of human-AI collaboration in VC decision-making and offers further practical insights for VC firms, developers and entrepreneurs.