AI Agents and Entrepreneurial Performance

Last registered on August 25, 2025
View Trial History

Pre-Trial

Trial Information

General Information

Title
AI Agents and Entrepreneurial Performance
RCT ID
AEARCTR-0016572
Initial registration date
August 22, 2025

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
August 25, 2025, 8:47 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country
United States of America
Region

Primary Investigator

Name
Jinkyong Choi
Affiliation
Columbia University
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Weishung Liu
PI Affiliation
Microsoft
Contact Investigator

Additional Trial Information

Status
In development
Start date
2025-08-11
End date
2025-09-09
Keywords
Firms & Productivity, Health, Labor
Additional Keywords
Entrepreneurship, Artificial Intelligence, Agents
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
We will test whether giving founder teams full access to AI agents helps them build better products faster and make stronger business decisions in a short, time‑boxed creation setting. Teams will be randomly assigned to either unlimited AI agent access or limited AI agent access during the event. Outcomes will focus on overall product progress, innovation, and business strategy quality. Data will come from surveys, evaluation scores, and anonymized usage patterns. Participation is voluntary, informed consent is collected, and all materials are handled confidentially. Our goal is to provide clear, non‑technical evidence on how AI agents affect startup teams in practice.
External Link(s)

Registration Citation

Citation
Choi, Jinkyong and Weishung Liu. 2025. "AI Agents and Entrepreneurial Performance." AEA RCT Registry. August 25. https://doi.org/10.1257/rct.16572-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
Teams of founders will be randomly assigned to either unlimited AI agent access or limited AI agent access during a short, time‑boxed software creation event. One group will work with full agentic capabilities, while the other will have only basic AI assistance. Access applies only during the event window, and outcomes focus on broad measures of product progress, innovation, and business strategy.
Intervention (Hidden)
Participants work in founder teams during a two‑day hackathon. Teams are randomly assigned to one of two conditions:

- Treatment: Full, unrestricted access to advanced AI agent capabilities.
- Control: Limited access to basic AI agent functionality.

The intervention is applied only during the event. Randomization occurs at the team level after formation. Outcomes will be assessed using surveys, evaluation scores, and anonymized usage data to understand how AI agents influence product development, innovation, and strategic decisions. All participation is voluntary, with informed consent and confidentiality protections in place.
Intervention Start Date
2025-09-08
Intervention End Date
2025-09-09

Primary Outcomes

Primary Outcomes (end points)
We will look at overall team performance and strategic quality during the event. Key outcomes include:
- Product Development Progress : How far teams get in building a working solution.
- Innovation : The originality and creativity of what teams produce.
- Business Strategy : The clarity and quality of the team’s go‑to‑market approach.
Primary Outcomes (explanation)
These outcomes will be constructed from multiple observable indicators:
- Product Development Progress: Based on advancement toward a working solution, drawing on evaluation scores and team self‑reports.
- Innovation: Reflects originality and creativity, assessed through expert and peer evaluations.
- Business Strategy: Captures clarity and quality of the team’s plan, using structured judgment criteria and participant feedback.

Secondary Outcomes

Secondary Outcomes (end points)
Agentic AI Adoption Decisions – Whether and how teams integrate AI agents into their workflow.
Secondary Outcomes (explanation)
Measured through anonymized usage logs and self‑reported adoption patterns during the event.

Experimental Design

Experimental Design
This study uses a randomized design to compare two conditions during a short, time‑boxed software creation event. Founder teams are randomly assigned to either full access to advanced AI agents or limited access to basic AI agent functionality. The goal is to estimate the causal effect of AI agents on team performance and strategic decision‑making. Outcomes will focus on overall product progress, innovation, and business strategy quality. Participation is voluntary, and all materials are handled confidentially.
Experimental Design Details
Randomization Method
Randomization was conducted by computer after team formation was complete. Teams were assigned to one of the two study conditions using a simple random draw to ensure fairness and eliminate selection bias.
Randomization Unit
Team
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
150 teams
Sample size: planned number of observations
150 teams
Sample size (or number of clusters) by treatment arms
75 teams control, 75 teams treatment
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Microsoft Research Institutional Review Board
IRB Approval Date
2025-08-13
IRB Approval Number
IRB 8974
Analysis Plan

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials