Dashboard

Texas A&M University — Mays Business School

AI VENTURE VELOCITY CHALLENGE
2026

This dashboard tracks how student venture teams across 160 organizations moved from application to evidence, from evidence to semifinalist selection, and eventually to the finals at Texas A&M.

From Applications to Evidence

Story thesis: A different kind of student venture competition

The AI Venture Velocity Challenge was designed to test a different model of student venture competition: not who could polish the best pitch, but who could learn fastest, test harder assumptions, and use AI to accelerate real venture progress.

  • 531 Applications Original applications received from student teams.
  • 248 Stage 1 Teams Teams submitted at least one formal experiment log.
  • 985 Formal Logs Google Form experiment-log submissions in the judge package.
  • 27 Semifinalists Teams advanced from Stage 1 to Stage 2.
  • 12FinalistsSelected for the September 26 finals.

Explore Stage 2: 210 submissions from the 27 semifinalists

The Story This Dashboard Tells

The Starting Hypothesis

Students using AI thoughtfully should be able to test ideas, understand customers, and make venture decisions faster than traditional student venture timelines allow.

The Behavior We Wanted

Teams were asked to move from claims to evidence: identify assumptions, run experiments, document learning, and show how the venture changed.

The Pattern We Are Watching

The strongest teams were not simply more active. They tended to leave clearer evidence trails around learning velocity, customer understanding, and disciplined execution.

National Reach The challenge opened with 531 applications from 160 institutions, creating a broad field of student venture teams.
Evidence at Scale Stage 1 produced 985 formal experiment logs from 248 student venture teams.
Selection Lens The rubric placed its largest weight on learning velocity and experimentation rigor, not pitch polish alone.

How the Challenge Works

May 1, 2026-September 26, 2026: Teams move from venture snapshots to Stage 1 experiment logs, semifinalist selection, finalist selection, and the finals at Texas A&M.

  1. 1Applications: Student teams submitted venture snapshots.
  2. 2Stage 1: Teams documented experiments, evidence, and learning.
  3. 3Semifinalists: Twenty-seven teams advanced to the next stage.
  4. 4Top 12 Finalists: Twelve finalists were announced September 3.
  5. 5Finals at Texas A&M: Finalist teams will compete in College Station in September.

Evaluation Rubric

  • Learning Velocity and Experimentation Rigor
    30%
  • Problem Significance and Opportunity Quality
    25%
  • AI Integration and Leverage
    20%
  • Evidence of Venture Progress
    10%
  • Responsible Impact and Ethical Design
    10%
  • Adaptive Execution and Coachability
    5%

Supported by

  • DeloitteFoundation
  • Deloitte
  • Midjourney

The largest weight was placed on learning velocity and experimentation rigor, reflecting the challenge’s core focus: helping student founders test, learn, and build faster with AI.

Applications

Chapter question: Who showed up?

The challenge began with 531 applications from student venture teams representing 160 institutions. The mapped U.S. records show broad national reach, a wide mix of industries, and many different approaches to building with AI.

What The Application Pool Showed

What We Expected

If the challenge spoke to a real shift in venture building, student teams from many schools and sectors would respond.

What The Data Showed

The first pool included 531 applications from 160 institutions, with mapped records spanning every U.S. census region.

What Stood Out

The field was not limited to one industry or one kind of AI use. Teams applied AI across health care, energy, education, workflows, public problems, and consumer needs.

  • 531 Original Applications Student venture applications received at launch.
  • 160 Institutions Schools represented in the full original application pool.
  • 438 Structured Snapshots Venture snapshots with structured stage and customer fields.
  • 248 Stage 1 Teams Teams that moved from application into documented experimentation.

Application Footprint

The mapped application source shows broad participation across every U.S. census region.

University of Arizona, Tucson, AZ: 1 team University of Arkansas at Pine Bluff, Pine Bluff, AR: 1 team California State University East Bay, Hayward, CA: 1 team California State University Los Angeles, Los Angeles, CA: 1 team University of California, San Diego, La Jolla, CA: 1 team University of California Santa Barbara, Santa Barbara, CA: 1 team Menlo College, Atherton, CA: 1 team San Diego State University, San Diego, CA: 1 team George Washington University, Washington, DC: 1 team Georgetown University, Washington, DC: 1 team Howard University, Washington, DC: 1 team University of Florida, Gainesville, FL: 1 team Polk State College, Winter Haven, FL: 1 team St. Thomas University, Miami Gardens, FL: 1 team University of Illinois Urbana Champaign, Champaign, IL: 1 team Southern Illinois University Edwardsville, Edwardsville, IL: 1 team Butler University, Indianapolis, IN: 1 team Indiana University Kelley Direct MBA, Bloomington, IN: 1 team Buena Vista University, Storm Lake, IA: 1 team Iowa State University, Ames, IA: 1 team Morningside University, Sioux City, IA: 1 team Pittsburg State University, Pittsburg, KS: 1 team Campbellsville University, Campbellsville, KY: 1 team University of Louisville, Louisville, KY: 1 team Midway University, Midway, KY: 1 team Tulane University, New Orleans, LA: 1 team Xavier University of Louisiana, New Orleans, LA: 1 team Loyola University of Maryland, Baltimore, MD: 1 team University of Maryland, College Park, College Park, MD: 1 team University of Maryland Eastern Shore, Princess Anne, MD: 1 team Morgan State University, Baltimore, MD: 1 team Prince George’s Community College, Largo, MD: 1 team Babson College, Wellesley, MA: 1 team Boston College, Chestnut Hill, MA: 1 team University of Michigan, Ann Arbor, MI: 1 team Wayne State University, Detroit, MI: 1 team Jackson State University, Jackson, MS: 1 team University of Missouri Kansas City, Kansas City, MO: 1 team Dartmouth College, Hanover, NH: 1 team Princeton University, Princeton, NJ: 1 team Eastern NM University, Portales, NM: 1 team Columbia University, New York, NY: 1 team Borough of Manhattan Community College, New York, NY: 1 team High Point University, High Point, NC: 1 team University of North Carolina Chapel Hill, Chapel Hill, NC: 1 team University of North Carolina Wilmington, Wilmington, NC: 1 team University of North Carolina at Pembroke, Pembroke, NC: 1 team Denison University, Granville, OH: 1 team Southeastern Oklahoma State University, Durant, OK: 1 team Robert Morris University, Moon Township, PA: 1 team Villanova University, Villanova, PA: 1 team University of Rhode Island, Kingston, RI: 1 team University of South Carolina, Columbia, SC: 1 team South Carolina State University, Orangeburg, SC: 1 team Dakota State University, Madison, SD: 1 team Belmont University, Nashville, TN: 1 team Southern Adventist University, Collegedale, TN: 1 team University of Tennessee, Knoxville, Knoxville, TN: 1 team Baylor College of Medicine, Houston, TX: 1 team Dallas Baptist University, Dallas, TX: 1 team University of Houston-Downtown, Houston, TX: 1 team Lamar University, Beaumont, TX: 1 team University of North Texas, Denton, TX: 1 team University of Texas Rio Grande Valley, Edinburg, TX: 1 team University of Texas at San Antonio, San Antonio, TX: 1 team Texas Southern University, Houston, TX: 1 team Snow College, Ephraim, UT: 1 team Green Rover College, Auburn, WA: 1 team Western Washington University, Bellingham, WA: 1 team Shepherd University, Shepherdstown, WV: 1 team University of Wisconsin – Madison, Madison, WI: 1 team South Louisiana Community College, Lafayette, LA: 1 team University of Arkansas, Fayetteville, AR: 2 teams University of California, Santa Cruz, Santa Cruz, CA: 2 teams University of Silicon Valley, San Jose, CA: 2 teams University of Southern California, Los Angeles, CA: 2 teams Florida Atlantic University, Boca Raton, FL: 2 teams Florida State University, Tallahassee, FL: 2 teams Emory University, Atlanta, GA: 2 teams Georgia Institute of Technology, Atlanta, GA: 2 teams Spelman College, Atlanta, GA: 2 teams DePaul University, Chicago, IL: 2 teams University of Notre Dame, Notre Dame, IN: 2 teams University of Iowa, Iowa City, IA: 2 teams Louisiana State University, Baton Rouge, LA: 2 teams Johns Hopkins University, Baltimore, MD: 2 teams University of Maryland Global Campus, Adelphi, MD: 2 teams Towson University, Towson, MD: 2 teams Boston University, Boston, MA: 2 teams Southern New Hampshire University, Manchester, NH: 2 teams University of New Hampshire, Durham, NH: 2 teams Stevens Institute of Technology, Hoboken, NJ: 2 teams New Mexico State University, Las Cruces, NM: 2 teams The University of Tulsa, Tulsa, OK: 2 teams Saint Joseph’s University, Philadelphia, PA: 2 teams Clemson University, Clemson, SC: 2 teams Piedmont Technical College, Greenwood, SC: 2 teams Texas A&M University-Corpus Christi, Corpus Christi, TX: 2 teams University of Houston, Houston, TX: 2 teams University of Texas at Arlington, Arlington, TX: 2 teams Texas Christian University, Fort Worth, TX: 2 teams Brigham Young University, Provo, UT: 2 teams Reynolds Community College, Richmond, VA: 2 teams University of Washington, Seattle, WA: 2 teams Stanford University, Stanford, CA: 2 teams Collin College, McKinney, TX: 2 teams Soka University of America, Aliso Viejo, CA: 2 teams University of Alabama, Tuscaloosa, AL: 3 teams California State University, Northridge, Northridge, CA: 3 teams California State University, Sacramento, Sacramento, CA: 3 teams University of California, Davis, Davis, CA: 3 teams University of California, Los Angeles, Los Angeles, CA: 3 teams Yale University, New Haven, CT: 3 teams Florida International University, Miami, FL: 3 teams University of Chicago, Chicago, IL: 3 teams University of Nebraska-Lincoln, Lincoln, NE: 3 teams New York University, New York, NY: 3 teams Duke University, Durham, NC: 3 teams Oral Roberts University, Tulsa, OK: 3 teams Drexel University, Philadelphia, PA: 3 teams University of Pennsylvania, Philadelphia, PA: 3 teams University of Texas at Tyler, Tyler, TX: 3 teams Keck Graduate Institute, Claremont, CA: 3 teams University of Hawaii at Manoa, Honolulu, HI: 4 teams University at Buffalo, Buffalo, NY: 4 teams Midwestern State University, Wichita Falls, TX: 4 teams Hampton University, Hampton, VA: 4 teams Purdue University, West Lafayette, IN: 4 teams University of Colorado Boulder, Boulder, CO: 5 teams University of Miami, Coral Gables, FL: 5 teams Michigan State University, East Lansing, MI: 5 teams Cornell University, Ithaca, NY: 5 teams University of Cincinnati, Cincinnati, OH: 5 teams George Mason University, Fairfax, VA: 5 teams Grambling State University, Grambling, LA: 6 teams Harvard University, Cambridge, MA: 6 teams Washington University in St. Louis, St. Louis, MO: 6 teams Vanderbilt University, Nashville, TN: 6 teams Texas State University, San Marcos, TX: 6 teams University of Texas at Austin, Austin, TX: 6 teams The Pennsylvania State University, University Park, PA: 7 teams Southern Methodist University, Dallas, TX: 7 teams University of California, Berkeley, Berkeley, CA: 8 teams Middle Tennessee State University, Murfreesboro, TN: 8 teams University of Texas at Dallas, Richardson, TX: 8 teams Northwestern University, Evanston, IL: 9 teams Virginia Tech, Blacksburg, VA: 9 teams Rice University, Houston, TX: 10 teams University of Connecticut, Storrs, CT: 12 teams New Jersey Institute of Technology, Newark, NJ: 12 teams Arizona State University, Tempe, AZ: 15 teams Texas A&M University, College Station, TX: 148 teams
Reading the counts: The headline application count is 531 from 160 institutions. The mapped institution tables represent 529 application records across 152 U.S. institutions available in the map source.

National Reach

Mapped Applications by U.S. Census Region
Region Mapped Applications Share Institutions Largest Source
South 323 61.1% 76 Texas A&M University
Northeast 77 14.6% 24 University of Connecticut
West 74 14.0% 28 Arizona State University
Midwest 55 10.4% 24 Northwestern University

Largest Application Sources

Top Institutions in the Original Application Map Source
Institution Location Applications Share
Texas A&M University College Station, TX 148 28.0%
Arizona State University Tempe, AZ 15 2.8%
University of Connecticut Storrs, CT 12 2.3%
New Jersey Institute of Technology Newark, NJ 12 2.3%
Rice University Houston, TX 10 1.9%
Northwestern University Evanston, IL 9 1.7%
Virginia Tech Blacksburg, VA 9 1.7%
University of California, Berkeley Berkeley, CA 8 1.5%
Middle Tennessee State University Murfreesboro, TN 8 1.5%
University of Texas at Dallas Richardson, TX 8 1.5%
The Pennsylvania State University University Park, PA 7 1.3%
Southern Methodist University Dallas, TX 7 1.3%
View additional application mix details

Industries

Classified Venture Snapshots by Industry
Industry Ventures Share
Healthcare 95 21.7%
Industrial/Energy 87 19.9%
Productivity/Workflow 60 13.7%
Education 55 12.6%
Personal/Lifestyle 49 11.2%
Media/Creative 27 6.2%
Public Sector/Civic 25 5.7%
Finance 22 5.0%
Frontier/Deep Tech 10 2.3%
Other 8 1.8%

What Powered the Ventures

Common AI Patterns in Classified Venture Snapshots
AI Pattern Mentions Share
LLM Applications 300 69.1%
Decision Support 191 44.0%
Data and Analytics 121 27.9%
Workflow Automation 94 21.7%
Computer Vision 92 21.2%
AI Agents 88 20.3%
Voice AI 48 11.1%
Marketplace 28 6.5%
Search and Retrieval 21 4.8%
Robotics 18 4.1%

AI-pattern categories were identified in 434 readable snapshots. Teams could reflect more than one AI pattern, so these shares are not intended to add to 100%.

Venture Stage

Application-Stage Venture Maturity
Stage Ventures Share
Brand-New Idea 144 32.9%
Prototype 143 32.6%
MVP 91 20.8%
Has Traction 60 13.7%

Target Customers

Primary Customer Type at Application
Customer Type Ventures Share
Enterprise 147 33.6%
Consumer 138 31.5%
Small and Midsize Business 97 22.1%
Mixed Customer Types 28 6.4%
Government 16 3.7%
Research 12 2.7%

Stage 1 Experiment Logs

Chapter question: Who kept testing and learning?

Stage 1 asked teams to test assumptions, gather evidence, use AI as a learning accelerator, and document meaningful progress.

From Participation To Evidence

What We Expected

Some teams would treat the challenge like a deadline. Others would treat it like a learning cycle and use the full window to test, decide, and iterate.

What The Data Showed

Two hundred forty-eight teams submitted formal experiment logs, creating 985 records of venture learning during Stage 1.

What Stood Out

Semifinalists were not selected by log count alone. But the teams that advanced often gave judges a clearer trail of evidence, decisions, and progress.

Stage 1 Footprint

Teams that submitted experiment logs continued to represent a broad national field.

Menlo College, Atherton, CA: 1 team South Louisiana Community College, Lafayette, LA: 1 team Tulane University, New Orleans, LA: 1 team Morgan State University, Baltimore, MD: 1 team Texas State University, San Marcos, TX: 1 team Boston College, Chestnut Hill, MA: 1 team University of Houston-Downtown, Houston, TX: 1 team Yale University, New Haven, CT: 1 team Stevens Institute of Technology, Hoboken, NJ: 1 team Johns Hopkins Bloomberg School of Public Health, Baltimore, MD: 1 team Florida State University, Tallahassee, FL: 1 team High Point University, High Point, NC: 1 team Emory University, Atlanta, GA: 1 team University of Southern California, Los Angeles, CA: 1 team San Diego State University, San Diego, CA: 1 team Stanford University, Stanford, CA: 1 team Carnegie Mellon University, Pittsburgh, PA: 1 team Georgetown University, Washington, DC: 1 team Western Washington University, Bellingham, WA: 1 team Robert Morris University, Moon Township, PA: 1 team Florida Atlantic University, Boca Raton, FL: 1 team Wayne State University, Detroit, MI: 1 team University of Wisconsin – Madison, Madison, WI: 1 team Washington University in St. Louis, St. Louis, MO: 1 team Polk State College, Winter Haven, FL: 1 team University of Houston, Houston, TX: 1 team University of North Carolina Chapel Hill, Chapel Hill, NC: 1 team Brigham Young University, Provo, UT: 1 team Southern New Hampshire University, Manchester, NH: 1 team University of Notre Dame, Notre Dame, IN: 1 team California State University, Sacramento, Sacramento, CA: 1 team University at Buffalo, Buffalo, NY: 1 team Pittsburg State University, Pittsburg, KS: 1 team Dallas Baptist University, Dallas, TX: 1 team University of Tennessee, Knoxville, Knoxville, TN: 1 team Drexel University, Philadelphia, PA: 1 team Clemson University, Clemson, SC: 1 team Prince George’s Community College, Largo, MD: 1 team Loyola University of Maryland, Baltimore, MD: 1 team Denison University, Granville, OH: 1 team Baylor College of Medicine, Houston, TX: 1 team University of Texas Rio Grande Valley, Edinburg, TX: 1 team California State University East Bay, Hayward, CA: 1 team Snow College, Ephraim, UT: 1 team Oral Roberts University, Tulsa, OK: 1 team University of Cincinnati, Cincinnati, OH: 1 team University of Iowa, Iowa City, IA: 1 team California State University, Northridge, Northridge, CA: 1 team Xavier University of Louisiana, New Orleans, LA: 1 team Iowa State University, Ames, IA: 1 team College of Marin, Kentfield, CA: 1 team University of Washington, Seattle, WA: 1 team University of Rhode Island, Kingston, RI: 1 team University of Arkansas at Pine Bluff, Pine Bluff, AR: 1 team Princeton University, Princeton, NJ: 1 team The University of Tulsa, Tulsa, OK: 1 team Saint Joseph’s University, Philadelphia, PA: 1 team Dakota State University, Madison, SD: 1 team Southeastern Oklahoma State University, Durant, OK: 1 team University of North Carolina Wilmington, Wilmington, NC: 1 team University of California, Davis, Davis, CA: 2 teams University of Chicago, Chicago, IL: 2 teams Florida International University, Miami, FL: 2 teams University of Miami, Coral Gables, FL: 2 teams Spelman College, Atlanta, GA: 2 teams New York University, New York, NY: 2 teams University of Texas at Tyler, Tyler, TX: 2 teams University of Alabama, Tuscaloosa, AL: 2 teams Hampton University, Hampton, VA: 2 teams Texas Christian University, Fort Worth, TX: 2 teams Keck Graduate Institute, Claremont, CA: 2 teams Harvard University, Cambridge, MA: 2 teams University of Texas at Arlington, Arlington, TX: 2 teams Rice University, Houston, TX: 2 teams University of Silicon Valley, San Jose, CA: 2 teams Boston University, Boston, MA: 2 teams University of Hawaii at Manoa, Honolulu, HI: 2 teams University of Maryland Global Campus, Adelphi, MD: 2 teams University of New Hampshire, Durham, NH: 2 teams Collin College, McKinney, TX: 2 teams University of California, Los Angeles, Los Angeles, CA: 2 teams Cornell University, Ithaca, NY: 3 teams Midwestern State University, Wichita Falls, TX: 3 teams University of Colorado Boulder, Boulder, CO: 3 teams Duke University, Durham, NC: 3 teams Purdue University, West Lafayette, IN: 3 teams Grambling State University, Grambling, LA: 3 teams Vanderbilt University, Nashville, TN: 3 teams Middle Tennessee State University, Murfreesboro, TN: 4 teams Southern Methodist University, Dallas, TX: 4 teams University of Texas at Austin, Austin, TX: 4 teams The Pennsylvania State University, University Park, PA: 4 teams Northwestern University, Evanston, IL: 4 teams University of Texas at Dallas, Richardson, TX: 4 teams George Mason University, Fairfax, VA: 5 teams University of California, Berkeley, Berkeley, CA: 5 teams Arizona State University, Tempe, AZ: 5 teams New Jersey Institute of Technology, Newark, NJ: 6 teams Virginia Tech, Blacksburg, VA: 7 teams University of Connecticut, Storrs, CT: 7 teams Texas A&M University, College Station, TX: 65 teams
View Stage 1 institutions represented on the map
Stage 1 Institutions by Formal Experiment Logs
Institution Location Teams Formal Logs Share of Formal Logs
Texas A&M University College Station, TX 65 252 25.6%
Virginia Tech Blacksburg, VA 7 49 5.0%
University of Connecticut Storrs, CT 7 35 3.6%
New Jersey Institute of Technology Newark, NJ 6 27 2.7%
Florida International University Miami, FL 2 26 2.6%
George Mason University Fairfax, VA 5 23 2.3%
University of Texas at Austin Austin, TX 4 19 1.9%
University of California, Berkeley Berkeley, CA 5 18 1.8%
Arizona State University Tempe, AZ 5 16 1.6%
Vanderbilt University Nashville, TN 3 16 1.6%
The Pennsylvania State University University Park, PA 4 15 1.5%
Duke University Durham, NC 3 15 1.5%
Northwestern University Evanston, IL 4 14 1.4%
University of Colorado Boulder Boulder, CO 3 14 1.4%
University of Chicago Chicago, IL 2 14 1.4%
University of Texas at Dallas Richardson, TX 4 13 1.3%
Southern Methodist University Dallas, TX 4 13 1.3%
Grambling State University Grambling, LA 3 12 1.2%
Cornell University Ithaca, NY 3 12 1.2%
Baylor College of Medicine Houston, TX 1 12 1.2%
Middle Tennessee State University Murfreesboro, TN 4 11 1.1%
Harvard University Cambridge, MA 2 11 1.1%
University of Silicon Valley San Jose, CA 2 10 1.0%
University of Miami Coral Gables, FL 2 10 1.0%
Collin College McKinney, TX 2 10 1.0%
High Point University High Point, NC 1 10 1.0%
University of Texas at Arlington Arlington, TX 2 9 0.9%
Spelman College Atlanta, GA 2 9 0.9%
University at Buffalo Buffalo, NY 1 9 0.9%
Denison University Granville, OH 1 9 0.9%
Purdue University West Lafayette, IN 3 8 0.8%
University of Alabama Tuscaloosa, AL 2 8 0.8%
Boston University Boston, MA 2 8 0.8%
Tulane University New Orleans, LA 1 8 0.8%
Florida State University Tallahassee, FL 1 8 0.8%
University of Texas at Tyler Tyler, TX 2 7 0.7%
University of New Hampshire Durham, NH 2 7 0.7%
Western Washington University Bellingham, WA 1 7 0.7%
The University of Tulsa Tulsa, OK 1 7 0.7%
Iowa State University Ames, IA 1 7 0.7%
Dakota State University Madison, SD 1 7 0.7%
University of Maryland Global Campus Adelphi, MD 2 6 0.6%
University of Hawaii at Manoa Honolulu, HI 2 6 0.6%
Rice University Houston, TX 2 6 0.6%
University of North Carolina Chapel Hill Chapel Hill, NC 1 6 0.6%
Saint Joseph’s University Philadelphia, PA 1 6 0.6%
Polk State College Winter Haven, FL 1 6 0.6%
Midwestern State University Wichita Falls, TX 3 5 0.5%
University of Iowa Iowa City, IA 1 5 0.5%
University of Tennessee, Knoxville Knoxville, TN 1 5 0.5%
Pittsburg State University Pittsburg, KS 1 5 0.5%
Texas Christian University Fort Worth, TX 2 4 0.4%
Keck Graduate Institute Claremont, CA 2 4 0.4%
Hampton University Hampton, VA 2 4 0.4%
Wayne State University Detroit, MI 1 4 0.4%
University of Wisconsin – Madison Madison, WI 1 4 0.4%
University of Washington Seattle, WA 1 4 0.4%
University of Texas Rio Grande Valley Edinburg, TX 1 4 0.4%
University of Houston-Downtown Houston, TX 1 4 0.4%
University of Arkansas at Pine Bluff Pine Bluff, AR 1 4 0.4%
Texas State University San Marcos, TX 1 4 0.4%
Southern New Hampshire University Manchester, NH 1 4 0.4%
Snow College Ephraim, UT 1 4 0.4%
Johns Hopkins Bloomberg School of Public Health Baltimore, MD 1 4 0.4%
University of California, Davis Davis, CA 2 3 0.3%
University of Southern California Los Angeles, CA 1 3 0.3%
University of Rhode Island Kingston, RI 1 3 0.3%
University of Notre Dame Notre Dame, IN 1 3 0.3%
Stevens Institute of Technology Hoboken, NJ 1 3 0.3%
South Louisiana Community College Lafayette, LA 1 3 0.3%
San Diego State University San Diego, CA 1 3 0.3%
Princeton University Princeton, NJ 1 3 0.3%
Prince George’s Community College Largo, MD 1 3 0.3%
Menlo College Atherton, CA 1 3 0.3%
Florida Atlantic University Boca Raton, FL 1 3 0.3%
Emory University Atlanta, GA 1 3 0.3%
Dallas Baptist University Dallas, TX 1 3 0.3%
College of Marin Kentfield, CA 1 3 0.3%
California State University, Sacramento Sacramento, CA 1 3 0.3%
California State University, Northridge Northridge, CA 1 3 0.3%
University of California, Los Angeles Los Angeles, CA 2 2 0.2%
New York University New York, NY 2 2 0.2%
Washington University in St. Louis St. Louis, MO 1 2 0.2%
Stanford University Stanford, CA 1 2 0.2%
Robert Morris University Moon Township, PA 1 2 0.2%
Oral Roberts University Tulsa, OK 1 2 0.2%
Georgetown University Washington, DC 1 2 0.2%
Drexel University Philadelphia, PA 1 2 0.2%
Clemson University Clemson, SC 1 2 0.2%
Yale University New Haven, CT 1 1 0.1%
Xavier University of Louisiana New Orleans, LA 1 1 0.1%
University of North Carolina Wilmington Wilmington, NC 1 1 0.1%
University of Houston Houston, TX 1 1 0.1%
University of Cincinnati Cincinnati, OH 1 1 0.1%
Southeastern Oklahoma State University Durant, OK 1 1 0.1%
Morgan State University Baltimore, MD 1 1 0.1%
Loyola University of Maryland Baltimore, MD 1 1 0.1%
Carnegie Mellon University Pittsburgh, PA 1 1 0.1%
California State University East Bay Hayward, CA 1 1 0.1%
Brigham Young University Provo, UT 1 1 0.1%
Boston College Chestnut Hill, MA 1 1 0.1%

Semifinalists Built a Longer Evidence Trail

The teams that advanced were not simply more active. They were more likely to document learning across multiple weeks, giving judges a clearer view of how assumptions, tests, decisions, and ventures changed over time.

  • Average Logs per Team 7.0 Semifinalists 3.6 Other Stage 1 teams Semifinalists documented almost twice as much formal learning.
  • Active in 4+ Weeks 48.1% Semifinalists 8.1% Other Stage 1 teams Semifinalists were far more likely to keep documenting progress across multiple weeks.
  • Median Evidence Span in Days 27 Semifinalists 1 Other Stage 1 teams The typical semifinalist left a longer trail between first and last log.

Full Stage 1 Comparison

Evidence Trail Metrics Across All Stage 1 Teams
Metric Semifinalists Other Stage 1 Teams
Avg. logs/team 7.0 3.6
Median active weeks 3 weeks 1 week
Active in 4+ weeks 48.1% 8.1%
Median first log date May 24 June 7
Median evidence span 27 days 1 day
Final-week log share 40.4% 60.5%

The median evidence span measures the time between a team’s first and last submitted experiment log. It does not mean other teams only worked for one day; it means their submitted evidence was more compressed near the end of Stage 1.

Consistency Check Among Active Teams

Same Metrics Among Teams With 4+ Formal Logs
Metric Semifinalists Other Teams
Avg. logs/team 7.2 5.8
Median active weeks 3.5 weeks 2.0 weeks
Active in 4+ weeks 50.0% 18.2%
Median first log date May 25 June 5
Median evidence span 24.7 days 13.9 days
Final-week log share 40.3% 61.1%

This view checks whether the pattern remains after excluding teams with fewer than 4 formal logs. The pattern still holds: semifinalist evidence was spread across more weeks and was less concentrated in the final deadline rush.

Log Submission Pattern

Formal Experiment Logs Per Team
Group Teams Average Median Maximum
All Stage 1 teams 248 3.97 4 25
Semifinalist teams 27 6.96 6 25
Other Stage 1 teams 221 3.61 3 16

Log volume is not a score. Judges evaluated the quality of evidence, relevance of experiments, and decisions teams made from what they learned.

Final Submission Rush

Teams documented experiments from May 4 through the June 20, 2026, deadline, with many submissions arriving in the final days.

Final 24 Hours
232 logs, 23.6%
Final 48 Hours
390 logs, 39.6%
Final 72 Hours
449 logs, 45.6%
Final Week
569 logs, 57.8%

What Teams Submitted

Most Common Experiment-Log Types Across All Stage 1 Teams
Experiment-Log Type Count Share
Customer discovery/interview learning 175 17.8%
Prototype or MVP build 132 13.4%
AI-enabled product or workflow test 131 13.3%
Technical feasibility test 89 9.0%
Prototype or MVP test with users 65 6.6%
Strategic pivot or major decision 61 6.2%
Market or competitor research 55 5.6%
General progress update with evidence 47 4.8%
Problem validation 46 4.7%
Testing AI agents for key business functions (e.g., product recommendations, customer communication) 34 3.5%

What Semifinalists Submitted

Most Common Experiment-Log Types Among Semifinalists
Experiment-Log Type Count Share
Customer discovery/interview learning 39 20.7%
AI-enabled product or workflow test 31 16.5%
Technical feasibility test 18 9.6%
Strategic pivot or major decision 16 8.5%
Prototype or MVP build 15 8.0%
Responsible AI/ethics/privacy/risk test 14 7.4%
Sales or customer acquisition test 10 5.3%
Pricing or business model test 8 4.3%
Market or competitor research 7 3.7%
Problem validation 6 3.2%

Documented Evidence Trail

Log count was not a score by itself. The grouped view shows how much formal evidence judges had available when assessing learning velocity, experiment rigor, and progress.

  • 1 Log
    Semifinalists 0.0%
    Other teams 23.1%
  • 2-3 Logs
    Semifinalists 3.7%
    Other teams 32.1%
  • 4-5 Logs
    Semifinalists 40.7%
    Other teams 26.7%
  • 6-8 Logs
    Semifinalists 29.6%
    Other teams 13.6%
  • 9+ Logs
    Semifinalists 25.9%
    Other teams 4.5%
Grouped Formal Logs Per Team
Logs Per Team Semifinalists Semifinalist Share Other Teams Other Share
1 Log 0 0.0% 51 23.1%
2-3 Logs 1 3.7% 71 32.1%
4-5 Logs 11 40.7% 59 26.7%
6-8 Logs 8 29.6% 30 13.6%
9+ Logs 7 25.9% 10 4.5%
Number of Formal Logs Submitted Per Team
Logs Per Team All Teams Semifinalists Other Stage 1 Teams
1 51 0 51
2 41 1 40
3 31 0 31
4 45 7 38
5 25 4 21
6 17 6 11
7 13 2 11
8 8 0 8
9 6 2 4
10 2 1 1
11 2 1 1
12 5 2 3
16 1 0 1
25 1 1 0

Industries Across the Funnel

The same broad industry buckets are used across stages to show how the field changed from applications to Stage 1 experiment logs and semifinalist selection.

  • Healthcare
    Applications 21.7%
    Stage 1 25.0%
    Semifinalists 29.6%
  • Industrial/Energy
    Applications 19.9%
    Stage 1 8.9%
    Semifinalists 7.4%
  • Productivity/Workflow
    Applications 13.7%
    Stage 1 37.5%
    Semifinalists 48.1%
  • Education
    Applications 12.6%
    Stage 1 8.5%
    Semifinalists 0.0%
  • Personal/Lifestyle
    Applications 11.2%
    Stage 1 8.5%
    Semifinalists 3.7%
  • Media/Creative
    Applications 6.2%
    Stage 1 4.0%
    Semifinalists 0.0%
  • Public Sector/Civic
    Applications 5.7%
    Stage 1 1.6%
    Semifinalists 3.7%
  • Finance
    Applications 5.0%
    Stage 1 2.0%
    Semifinalists 0.0%
  • Frontier/Deep Tech
    Applications 2.3%
    Stage 1 2.8%
    Semifinalists 7.4%
  • Other
    Applications 1.8%
    Stage 1 1.2%
    Semifinalists 0.0%
Industry Representation by Challenge Stage
Industry Applications Application Share Stage 1 Teams Stage 1 Share Semifinalists Semifinalist Share
Healthcare 95 21.7% 62 25.0% 8 29.6%
Industrial/Energy 87 19.9% 22 8.9% 2 7.4%
Productivity/Workflow 60 13.7% 93 37.5% 13 48.1%
Education 55 12.6% 21 8.5% 0 0.0%
Personal/Lifestyle 49 11.2% 21 8.5% 1 3.7%
Media/Creative 27 6.2% 10 4.0% 0 0.0%
Public Sector/Civic 25 5.7% 4 1.6% 1 3.7%
Finance 22 5.0% 5 2.0% 0 0.0%
Frontier/Deep Tech 10 2.3% 7 2.8% 2 7.4%
Other 8 1.8% 3 1.2% 0 0.0%

Industry categories are grouped for public reporting and use the same taxonomy across stages.

Semifinalists

Chapter question: What changed as the field narrowed?

The semifinalist pool reflects the challenge rubric: problem significance, learning velocity, experimentation rigor, AI integration, evidence of venture progress, responsible impact, and adaptive execution.

What Advancement Signaled

What We Expected

The strongest teams would combine meaningful problems with better evidence, not just stronger storytelling or more mature starting points.

What The Data Showed

Twenty-seven teams advanced from Stage 1 after judges reviewed their original snapshots, experiment logs, supporting evidence, and documented progress.

What Stood Out

The semifinalist group remained geographically distributed, with Texas A&M representation slightly lower than its share of the original application pool.

Semifinalist Footprint

The semifinalist pool remained geographically distributed while narrowing to teams with stronger evidence of learning velocity and venture progress.

Purdue University, West Lafayette, IN: 1 team The Pennsylvania State University, University Park, PA: 1 team Baylor College of Medicine, Houston, TX: 1 team University of Texas at Austin, Austin, TX: 1 team The University of Tulsa, Tulsa, OK: 1 team Rice University, Houston, TX: 1 team Cornell University, Ithaca, NY: 1 team University of North Carolina Chapel Hill, Chapel Hill, NC: 1 team Florida International University, Miami, FL: 1 team High Point University, High Point, NC: 1 team Pittsburg State University, Pittsburg, KS: 1 team George Mason University, Fairfax, VA: 1 team University of Silicon Valley, San Jose, CA: 1 team University of Wisconsin – Madison, Madison, WI: 1 team Southern Methodist University, Dallas, TX: 1 team University of Connecticut, Storrs, CT: 1 team Virginia Tech, Blacksburg, VA: 2 teams University of California, Berkeley, Berkeley, CA: 2 teams Texas A&M University, College Station, TX: 7 teams

Institution Representation

Texas A&M represented 25.9% of semifinalist teams, compared with 26.2% of Stage 1 log teams and 27.9% of the full application pool.

Semifinalist Institutions Compared With Earlier Stages
Institution Semifinalists Semifinalist Share Stage 1 Teams Stage 1 Share Applications Application Share
Texas A&M University 7 25.9% 65 26.2% 148 27.9%
Virginia Tech 2 7.4% 7 2.8% 9 1.7%
University of California, Berkeley 2 7.4% 5 2.0% 8 1.5%
Purdue University 1 3.7% 3 1.2% 4 0.8%
The Pennsylvania State University 1 3.7% 4 1.6% 7 1.3%
Baylor College of Medicine 1 3.7% 1 0.4% 1 0.2%
University of Texas at Austin 1 3.7% 1 0.4% 6 1.1%
The University of Tulsa 1 3.7% 1 0.4% 2 0.4%
Rice University 1 3.7% 2 0.8% 10 1.9%
Cornell University 1 3.7% 3 1.2% 5 0.9%
University of North Carolina Chapel Hill 1 3.7% 2 0.8% 1 0.2%
Florida International University 1 3.7% 2 0.8% 3 0.6%
High Point University 1 3.7% 1 0.4% 1 0.2%
Pittsburg State University 1 3.7% 1 0.4% 1 0.2%
George Mason University 1 3.7% 5 2.0% 5 0.9%
University of Silicon Valley 1 3.7% 2 0.8% 2 0.4%
University of Wisconsin – Madison 1 3.7% 1 0.4% 1 0.2%
Southern Methodist University 1 3.7% 4 1.6% 7 1.3%
University of Connecticut 1 3.7% 7 2.8% 12 2.3%

Institution names are normalized for comparison. Multi-institution teams can create small differences between public roster counts and primary-institution counts.

Semifinalist Evidence Varied

Across the semifinalist group, teams tended to document more learning over more time. But that pattern was not a rule. Judges advanced teams based on the quality of evidence, learning, venture progress, and disciplined decisions.

Semifinalist Counterexamples

  • 2 logs A team advanced with the fewest formal logs among semifinalists.
  • 5 logs in one day A team advanced after submitting all formal logs on the June 20 deadline.

Non-Advancing Counterexamples

  • 16 logs A non-advancing team submitted more formal logs than most semifinalists.
  • 12 logs over 46 days A non-advancing team documented evidence across nearly the full Stage 1 window.

The takeaway: more frequent, more sustained documentation was common among semifinalists, but activity alone did not determine advancement.

Student Level

Student Level Among Stage 1 Teams
LevelTeamsShare
Undergraduate Student 127 51.2%
Graduate Student 117 47.2%
Community College/Technical College Student 2 0.8%
Not specified 2 0.8%

Semifinalist Student Level

Student Level Among Semifinalists
LevelTeamsShare
Graduate Student 14 51.9%
Undergraduate Student 13 48.1%

AI Experience

AI Experience Among Stage 1 Teams
Experience LevelTeamsShare
Building With AI 123 49.6%
Advanced Users 46 18.5%
Regular Users 40 16.1%
Mixed Across The Team 24 9.7%
Occasional Users 9 3.6%
Beginner 4 1.6%
Not specified 2 0.8%

Semifinalist AI Experience

AI Experience Among Semifinalists
Experience LevelTeamsShare
Building With AI 18 66.7%
Mixed Across The Team 4 14.8%
Regular Users 2 7.4%
Advanced Users 1 3.7%
Occasional Users 1 3.7%
Beginner 1 3.7%
Interpretation: Advancement did not simply reward polish, school affiliation, or venture maturity. The evidence suggests the semifinalist group more often combined substantive problem spaces with clearer learning trails, stronger experiment discipline, and better use of AI to accelerate testing and decision-making.

Meet the Semifinalist Teams

These 27 teams advanced from Stage 1 after judges reviewed venture snapshots, experiment logs, supporting evidence, and documented progress against the published rubric.

AeroGraphiX

Advanced materials/Filtration/Clean air/Infection control
University/College
University of Connecticut | College of Liberal Arts and Sciences
Team
Deep Shikha Srivastava; Douglas Adamson; Robert Williams; Brenden Ferland

AeroGraphiX is developing graphene-modified HEPA filters designed to improve filtration and antiviral performance. The team stood out by pairing advanced materials research with AI-enabled market learning, using customer discovery to redirect the venture toward more specific, higher-need environments.

Atra

Sports operations/Athletics technology
University/College
Pittsburg State University | Crossland College of Technology
Team
Leo Chauchard; Baptiste D’Hondt; Edouard Meurant

Atra is an operations platform for NCAA track and field programs, helping coaches manage rosters, training, equipment, travel, and meet logistics. The team stood out by learning from real athletic environments, accepting where user behavior challenged its assumptions, and refining the adoption path around coaches.

BusFactor

Enterprise knowledge/Workforce risk/M&A integration
University/College
Purdue University | College of Engineering
Team
Tayseer Abdeljaber

BusFactor helps organizations identify critical knowledge gaps before employee transitions put operations at risk. The team stood out in Stage 1 by moving beyond a clever analytics concept, pressure-testing the problem with business and technical stakeholders, and refining the venture around a sharper path to adoption.

ChelysSecurity

Cybersecurity/Autonomous penetration testing
University/College
Florida International University | College of Engineering & Computing
Team
Alejandro Almeida; Jibram Jimenez; Joel Dos Santos; Lucas Cox

ChelysSecurity is developing adaptive AI agents for autonomous penetration testing with safety controls and evidence-based guardrails. The team stood out by treating responsible autonomy as a design challenge, using practitioner feedback to refine how human oversight, scope, and trust should shape the product.

Concept Bytes

Operating-room operations/Medtech procurement/Healthcare workflow
University/College
University of North Carolina at Chapel Hill | School of Education
Team
Shelby McCormick; William Andrew (Drew) Wyatt

Concept Bytes turns paper-based surgical preference cards into digital intelligence for operating-room teams. The team advanced by listening deeply to frontline professionals, discovering a larger knowledge-transfer problem, and reshaping the venture around the workflows that matter inside the OR.

Dhi

Computer vision/Safety analytics/Infrastructure monitoring
University/College
Rice University | College of Engineering
Team
Dev Sanghvi; Ansh Dabral; Madhuvani Thatiparti; Venkata Sai Aneesh Thatiparti

Dhi turns existing cameras into edge-based safety agents for real-time detection, monitoring, and infrastructure intelligence. The team stood out by using AI to accelerate both product development and customer outreach, then converting that speed into real discovery, demos, and market feedback.

Do Si

Drug discovery/Biotech/Pharmaceutical R&D
University/College
The University of Tulsa | College of Engineering & Computer Science; Tulane University | School of Science & Engineering
Team
Vusal Karimov; Toghrul Azizli

Do Si is developing a failure-aware AI platform to help identify toxic molecules earlier in drug development. The team advanced by using expert feedback to challenge its own technical assumptions and reshape the product around the trust, explainability, and workflow needs of deep-tech users.

EntroPINN

Industrial process optimization/Energy efficiency/Engineering software
University/College
University of California, Berkeley | College of Chemistry; College of Computing, Data Science, and Society
Team
Nathan Liu; Dongha (David) Kim; Christopher Donahue

EntroPINN is building a physics-informed AI copilot to help process engineers simulate, debug, and optimize industrial systems. The team advanced by combining serious technical depth with extensive customer discovery, grounding its AI work in the daily friction and judgment of engineering users.

HumaniCore AI

Responsible AI/HR tech/Compliance
University/College
The Pennsylvania State University | Smeal College of Business
Team
Haley Friary-Schenk; Aly Orloff; Rich Avila; Craig Gutjahr; Anirudh Badia; Dennis Hollenbeck

HumaniCore AI is building a certification and governance platform for AI-driven employment decisions. The team advanced by treating trust, validation, and responsible AI as core product requirements, using Stage 1 to test both the technical system and the credibility needed for adoption.

InfoSavvy

Higher education/Academic advising/Student success
University/College
Southern Methodist University | Lyle School of Engineering
Team
Yaw Boateng; Ameen Zia

InfoSavvy is an AI-powered assistant that helps students access academic advising, campus resources, events, and opportunities. The team stood out by testing real student demand during moments of advising need, using usage behavior and university feedback to refine the product beyond a generic chatbot.

Interlinked

Climate resilience/Wildfire intelligence/Emergency response
University/College
University of California, Berkeley | College of Letters and Science
Team
Rehaan Bicha; Neel Shetty; Rushil Desai

Interlinked is building an AI wildfire intelligence platform for risk prediction, response planning, and emergency decision-making. The team advanced by moving beyond a simple dashboard idea toward the data infrastructure and decision workflows needed by utilities, emergency teams, and resilience partners.

IntuSense

Medtech/ICU infection detection/Hospital-acquired infections
University/College
Texas A&M University | College of Engineering
Team
Idris Hussain; Dunebari Mii; Caroline Behnke

IntuSense is developing a sensor-based approach to help detect early signs of ventilator-associated pneumonia. The team advanced by using expert input to question its technical path, compare alternatives, and pivot responsibly toward a more promising detection strategy.

Lobe AI

Senior living/AI sales operations/Real estate customer acquisition
University/College
Texas A&M University | College of Engineering
Team
Tiernan Lindauer; Shubh Bhakta

Lobe AI is an AI web setter for senior living communities that talks with families, books tours, and gives sales teams better context. The team stood out for fast AI-native execution, testing both the product experience and the sales workflow needed to make the tool useful for operators.

MottoNote

Enterprise knowledge/Manufacturing AI/Knowledge graphs
University/College
University of Wisconsin – Madison | College of Computing and Artificial Intelligence
Team
Siddharth Singh; Viacheslav Iudenko

MottoNote turns fragmented company knowledge into a structured knowledge graph that AI can reason over. The team stood out by narrowing a broad organizational-memory idea toward industrial settings, where fragmented knowledge creates immediate operational friction and clearer value.

NEXUS

Clinical trials/Oncology/Health AI
University/College
Baylor College of Medicine; Columbia University
Team
Shay Beheshti; Brooke Dirvin

NEXUS is developing an AI-powered system to help match cancer patients with relevant clinical trials more efficiently. The team stood out by combining technical prototyping with clinical validation, testing whether its system could handle the complexity and precision required in oncology workflows.

Oncera Health

Oncology survivorship/Digital health
University/College
University of Texas at Austin | College of Natural Sciences
Team
Rutu Ruparel

Oncera Health translates clinical guidance into personalized daily actions for cancer survivors after treatment ends. In Stage 1, the team used stakeholder discovery to refine how survivorship support could move from a meaningful patient need toward a more credible healthcare adoption path.

OsageMD

Legal-tech/Healthcare finance/Medical lien operations
University/College
Texas A&M University | Mays Business School
Team
Matthew Schaefer

OsageMD automates medical lien case management for personal-injury law firms and medical providers. The team advanced by bringing AI into a messy, document-heavy operational workflow, testing real records and process constraints rather than relying on a polished concept alone.

Parkevo

Smart mobility/Parking analytics/Computer vision
University/College
High Point University | Earl N. Phillips School of Business
Team
Evan Taylor; Brianna Stinespring

Parkevo uses edge AI vision to give drivers real-time parking availability and administrators better utilization insights. The team advanced by turning a familiar campus frustration into a broader mobility opportunity, testing stakeholder needs, technical constraints, and expansion potential.

RaizenAI

Hiring/Talent assessment/Developer platforms
University/College
George Mason University | College of Engineering and Computing
Team
Jagan Yetukuri; Abey Paul; Priya Mohan

RaizenAI replaces resume-first developer screening with proof-of-work hiring based on real projects and AI-assisted skill evaluation. The team advanced by testing the core belief behind the business, comparing traditional hiring signals with performance evidence and refining its approach based on what the data showed.

Rally Dynamics

Sports robotics/Training technology/Computer vision
University/College
Texas A&M University | College of Engineering
Team
Vishanth Ramamurthy Yoganand; Ryan Daly

Rally Dynamics is building an AI-powered badminton training robot for academies and competitive players. The team advanced by pairing robotics development with feedback from players and academy owners, using that learning to refine the business model, training use cases, and path to adoption.

RELAI

Hospital operations/Care coordination/Health systems
University/College
Cornell University | SC Johnson College of Business; Weill Cornell Medicine | Graduate School of Medical Sciences
Team
Vijay Raghunathan; Galina Borodulina; Alexandra (Aly) Abrams; Gustavo Avila Amat; Carly Skudin

RELAI is building a real-time hospital transfer coordination platform for critically ill patients who need specialized care. The team stood out by bringing business and medical perspectives together, using stakeholder discovery and workflow testing to refine a high-stakes healthcare operations problem.

Reliat

Industrial AI/Mining operations/Predictive maintenance
University/College
Texas A&M University | College of Engineering
Team
Alice Queiroz; Phuong (Alex) Dao

Reliat brings AI into the physical backbone of mining and aggregate operations, turning industrial monitoring data into faster, clearer maintenance decisions. The team stood out in Stage 1 by getting close to real operators, pressure-testing its assumptions in the field, and refining the venture around the problems that drive downtime, risk, and lost productivity.

SecRecon

Cybersecurity/Managed security services
University/College
Virginia Tech | Pamplin Business School
Team
Lokesh Lalwani

SecRecon uses agentic AI to verify vulnerabilities and turn noisy security scans into evidence teams can act on. The team stood out by focusing on trust and verification, not just automation, and by using security-stakeholder discovery to sharpen the product around false positives and triage burden.

SmartSwine

Agtech/Precision livestock/Animal health
University/College
Texas A&M University | College of Agriculture & Life Sciences
Team
Ziyuan Zhao; Yu Wang; Emmanuel Otchere; Rolando Alaniz

SmartSwine uses precision livestock AI to help reduce piglet preweaning mortality in farrowing environments. The team advanced by combining agricultural research, producer feedback, and field-oriented product development, showing how AI can support practical animal-health decisions on the farm.

Vitalia AI

Consumer health/Chronic illness/Quantified-self tools
University/College
Virginia Tech | Pamplin College of Business
Team
Will Boyer

Vitalia AI helps people track health patterns across symptoms, habits, and daily data. The team stood out by listening to chronic-illness and quantified-self communities, using user feedback to move away from weaker use cases and focus more tightly on urgent health frustrations.

VoiceVault

Digital health/Assistive communication/Voice AI
University/College
Texas A&M University | College of Engineering
Team
Zarik Khan

VoiceVault helps patients facing permanent voice loss preserve or recreate a personal voice for assistive communication. The team distinguished itself by pairing AI voice technology with clinician and user feedback, using real-world needs to guide which patients and workflows the product should serve first.

Zero1 IO

Enterprise AI agents / Operations automation
University/College
University of Silicon Valley
Team
Aisha Medina; Suman Dangol

Zero1 IO deploys private vertical AI agents that turn messy operational work into ready-to-review drafts for complex businesses. The team stood out by learning that enterprise AI adoption requires focus, narrowing from a broad platform idea toward specific workflows where customers feel the pain immediately.

STAGE 2 / JULY-AUGUST 2026

27 ventures. 210 logs. 12 finalists.

The field narrowed. The work did not stop. The semifinalists continued testing ideas in the market and documenting what they learned.

27Teams with Stage 2 logsAll semifinalists represented
210Formal submissions94 finalist logs + 116 other semifinalist logs
12Finalists selected15 semifinalists did not advance
Sept. 26Live finalsPrize decisions are still ahead
Finalists (12 teams)Other semifinalists (15 teams)

More logs did not separate the groups

Finalists averaged 7.83 logs per team; the other semifinalists averaged 7.73. Their ranges overlapped substantially. Submission volume alone does not explain the selected cohort.

Stage 2 submission activity
MeasureFinalists (n=12)Other semifinalists (n=15)
Teams1215
Formal submitted logs94116
Average logs per team7.837.73
Median logs per team7.56
Range of logs per team3-173-22
Median distinct submission dates4.55
Teams submitting on a single date0/123/15
Logs dated July21/94 (22.3%)32/116 (27.6%)
Logs dated Aug. 19-2645/94 (47.9%)50/116 (43.1%)

These are form submissions, not audited counts of distinct experiments. A log may summarize several tests or repeat earlier evidence. Submission dates do not establish when the work occurred.

When the records arrived

Submission dates as recorded in the reviewed export: July 7-August 26.

July 7August 1August 26

Includes all 210 accepted submissions in the August 26 judging package. Five entries recorded just after midnight on August 26 were accepted by the organizers and are included throughout.

What the teams reported

These signals describe the record. They are not a reconstructed judging formula or independently verified business outcomes.

SignalFinalistsOther semifinalists
At least one log reporting product or service users12/12 (100%)12/15 (80%)
At least one log reporting external validation of consequential AI output11/12 (91.7%)14/15 (93.3%)
At least one standard response reporting a weakened or changed assumption10/12 (83.3%)11/15 (73.3%)

Users are not necessarily paying customers. The question includes free users, pilots and tests.

Validation was common in both groups. Reporting one validation step does not measure its rigor or frequency. These observations do not establish why a particular team advanced.

Each team is counted once for a reported signal. Missing answers are not treated as “No”; nonstandard outcome responses are excluded.

The problems they are working on

Health and biotech account for 4 of the 12 finalists, followed by industrial and infrastructure ventures (3), business workflow (3), agriculture (1) and mobility (1).

CategoryFinalistsOther semifinalistsShare of each cohort
Agriculture1/12 (8.3%)0/15 (0.0%)
Business workflow3/12 (25.0%)2/15 (13.3%)
Climate / civic0/12 (0.0%)1/15 (6.7%)
Education0/12 (0.0%)1/15 (6.7%)
Health / biotech4/12 (33.3%)4/15 (26.7%)
Industrial / infrastructure3/12 (25.0%)2/15 (13.3%)
Mobility1/12 (8.3%)0/15 (0.0%)
Security / responsible AI0/12 (0.0%)3/15 (20.0%)
Sports technology0/12 (0.0%)2/15 (13.3%)

Each venture is grouped by its primary customer problem, not the AI technology it uses. The same rule applies to both cohorts. Categories describe the field, not venture quality or a sector preference, and differ from the original application categories.

How industries are grouped

Health / biotech: care, clinical operations, assistive health technology and drug research. Business workflow: organizational knowledge, sales, hiring and financial or administrative workflows, including Lobe AI and OsageMD. Industrial / infrastructure: manufacturing, mining, engineering simulation, materials and infrastructure monitoring, including ZERO1 IO, Dhi and AeroGraphiX.

The remaining groups are Security / responsible AI, Education, Agriculture, Mobility, Sports technology and Climate / civic. Detailed venture descriptions retain the context that a single category cannot capture.

Academic backgrounds, not founder ages

CategoryFinalistsOther semifinalistsShare of each cohort
Graduate5/12 (41.7%)9/15 (60.0%)
Undergraduate7/12 (58.3%)6/15 (40.0%)

Counts are teams, classified by the primary contact’s student level. Finalist profiles use the corrected public roster; other semifinalists use reconciled confirmation records. Mixed teams can include other levels. Founder ages are not available on a consistent basis and are not inferred.

Where the teams come from

Texas A&M is the primary institution for 5 of 12 finalists and 2 of 15 other semifinalists. The finalist roster represents 10 institutions when secondary affiliations are included.

Compare primary institutions across both cohorts
CategoryFinalistsOther semifinalistsShare of each cohort
Baylor College of Medicine1/12 (8.3%)0/15 (0.0%)
Cornell University0/12 (0.0%)1/15 (6.7%)
Florida International University0/12 (0.0%)1/15 (6.7%)
George Mason University0/12 (0.0%)1/15 (6.7%)
High Point University1/12 (8.3%)0/15 (0.0%)
Pittsburg State University0/12 (0.0%)1/15 (6.7%)
Purdue University0/12 (0.0%)1/15 (6.7%)
Rice University0/12 (0.0%)1/15 (6.7%)
Southern Methodist University0/12 (0.0%)1/15 (6.7%)
Texas A&M University5/12 (41.7%)2/15 (13.3%)
The Pennsylvania State University0/12 (0.0%)1/15 (6.7%)
The University of North Carolina Chapel Hill0/12 (0.0%)1/15 (6.7%)
The University of Tulsa1/12 (8.3%)0/15 (0.0%)
University of California, Berkeley1/12 (8.3%)1/15 (6.7%)
University of Connecticut0/12 (0.0%)1/15 (6.7%)
University of Silicon Valley1/12 (8.3%)0/15 (0.0%)
University of Texas at Austin1/12 (8.3%)0/15 (0.0%)
University of Wisconsin-Madison1/12 (8.3%)0/15 (0.0%)
Virginia Tech0/12 (0.0%)2/15 (13.3%)

One primary institution per team in this comparison. Secondary affiliations appear on finalist profiles. Institution counts describe representation; they do not establish a selection preference or a fairness conclusion.

What we can say now

Activity is not the same as progress

The average log counts are almost identical. The next layer of understanding is about what teams tested and what their evidence justified.

Market contact was not exclusive to finalists

All finalists reported users in at least one cycle, but so did most of the remaining semifinalists. The depth of validation matters.

The finalist field is not uniformly mature

Undergraduate and graduate founders are represented, alongside ventures at different development stages. The opportunity to win is not reserved for a single founder profile.

Small, selected cohorts; descriptive associations only. No causal claim about AI, demographics, industry or selection follows from these comparisons.

Sources and scope

Stage 2 counts reconcile to the August 26 judging package: 27 teams, 210 indexed records, no duplicate team/timestamp/title keys. Selection and corrected finalist affiliations follow the September 3 finalist announcement.

Only aggregate responses and public-facing venture information are included. Private log narratives, contact details, supporting-evidence links and individual judge ballots are not included. Meet the judging community on the judges page.

SELECTED SEPTEMBER 2026 / ALPHABETICAL, NOT RANKED

Meet the 12 finalists

Different problems. Different starting points. A shared opportunity to demonstrate evidence-backed venture progress in College Station.

7 / 5Undergraduate / graduateTeam-level academic classification
10Institutions represented8 primary institutions + 2 secondary affiliations
24Listed team membersAcross 12 ventures; 3 solo-member teams

Finalist Footprint

Twelve ventures represent ten institutions across the country.

Texas A&M University, College Station, TX: 5 finalist teams (Lobe AI, OsageMD, Reliat, Smart Swine, SpeechRoot)Baylor College of Medicine, Houston, TX: 1 finalist team (NEXUS)Columbia University, New York, NY: 1 finalist team (NEXUS)High Point University, High Point, NC: 1 finalist team (Parkevo)The University of Tulsa, Tulsa, OK: 1 finalist team (Do Si)Tulane University, New Orleans, LA: 1 finalist team (Do Si)University of California, Berkeley, Berkeley, CA: 1 finalist team (EntroPINN)University of Silicon Valley, San Jose, CA: 1 finalist team (ZERO1 IO)University of Texas at Austin, Austin, TX: 1 finalist team (Oncera Health)University of Wisconsin – Madison, Madison, WI: 1 finalist team (MottoNote)

Circle size represents finalist teams affiliated with each institution. Includes secondary affiliations: Do Si represents Tulsa and Tulane; NEXUS represents Baylor College of Medicine and Columbia. A team may appear at more than one location.

Health / biotech

Do Si

Earlier identification of potentially toxic drug candidates.

Learning in Stage 2: Technical scrutiny helped narrow the product around interpretable results and pharmaceutical workflows.

University
The University of Tulsa; Tulane University
College / school
McDougall School of Petroleum Engineering; School of Science & Engineering
Academic profile
Graduate | Petroleum; chemical & biomolecular engineering
Team
2 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Functional predictive-model prototype; generative interface still to come.

Stage 2: 9 formal submissions across 8 dates. These counts are descriptive, not scores.

Industrial / infrastructure

EntroPINN

An engineering copilot for more efficient industrial simulations.

Learning in Stage 2: Benchmarking and industry discovery led the team to narrow the scope of its model.

University
University of California, Berkeley
College / school
College of Chemistry; College of Computing, Data Science, and Society
Academic profile
Undergraduate | Chemical engineering; computer science
Team
3 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Working simulation copilot; pre-revenue and pre-commercial launch.

Stage 2: 8 formal submissions across 5 dates. These counts are descriptive, not scores.

Business workflow

Lobe AI

AI-assisted sales follow-up for senior living communities.

Learning in Stage 2: Customer discovery progressed into a paid engagement and additional customer commitments.

University
Texas A&M University
College / school
College of Engineering; Mays Business School
Academic profile
Undergraduate | Computer science, physics; finance
Team
2 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: MVP in operator feedback testing; live family-facing deployment planned.

Stage 2: 4 formal submissions across 3 dates. These counts are descriptive, not scores.

Business workflow

MottoNote

Organizational knowledge that stays useful as information changes.

Learning in Stage 2: Customer usage, rather than a feature request alone, guided product priorities.

University
University of Wisconsin-Madison
College / school
College of Computing and Artificial Intelligence
Academic profile
Undergraduate | Computer science
Team
2 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Knowledge-compiler product; pre-revenue; pilot discussions underway.

Stage 2: 5 formal submissions across 4 dates. These counts are descriptive, not scores.

Health / biotech

NEXUS

Matching cancer patients with suitable clinical trials.

Learning in Stage 2: A pilot commitment sharpened attention to clinical integration and governance.

University
Baylor College of Medicine; Columbia University
College / school
Genetics and genomics; business development (college not specified)
Academic profile
Graduate | Genetics/genomics; business development
Team
2 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Pre-MVP proof of concept on synthetic data; no production deployment.

Stage 2: 9 formal submissions across 6 dates. These counts are descriptive, not scores.

Health / biotech

Oncera Health

Personalized support for people recovering after cancer treatment.

Learning in Stage 2: Patient needs were examined alongside reimbursement and health-system adoption.

University
University of Texas at Austin
College / school
College of Natural Sciences
Academic profile
Undergraduate | Molecular & cellular biology
Team
1 listed member
Venture starting point and submissions

Original snapshot at the start of the challenge: Functional platform available for pilot use; reported IRB-approved pilot.

Stage 2: 9 formal submissions across 9 dates. These counts are descriptive, not scores.

Business workflow

OsageMD

Case management for medical-lien financing operations.

Learning in Stage 2: Workflow research and technical testing informed conversations with economic buyers.

University
Texas A&M University
College / school
Mays Business School
Academic profile
Graduate | Business administration
Team
1 listed member
Venture starting point and submissions

Original snapshot at the start of the challenge: Platform already in production with a launch operator (reported 8 months).

Stage 2: 3 formal submissions across 2 dates. These counts are descriptive, not scores.

Mobility

Parkevo

Live parking information from edge-based computer vision.

Learning in Stage 2: A signed pilot and deployment work brought institutional discovery closer to market use.

University
High Point University
College / school
Earl N. Phillips School of Business
Academic profile
Undergraduate | Finance; social media & digital communication
Team
2 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Prelaunch, pilot-ready; site and partners identified; deployment ahead.

Stage 2: 17 formal submissions across 15 dates. These counts are descriptive, not scores.

Industrial / infrastructure

Reliat

AI intelligence for mining and aggregate operations.

Learning in Stage 2: Industry discovery narrowed the initial concept and reshaped the product and commercial approach.

University
Texas A&M University
College / school
College of Engineering
Academic profile
Undergraduate | Electrical engineering; data engineering
Team
2 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Software MVP complete; pre-revenue; hardware MVP was the next step.

Stage 2: 13 formal submissions across 3 dates. These counts are descriptive, not scores.

Agriculture

Smart Swine

Livestock monitoring to improve piglet survival.

Learning in Stage 2: Commercial-farm testing connected research with producer economics and pricing questions.

University
Texas A&M University
College / school
College of Agriculture and Life Sciences; College of Engineering
Academic profile
Graduate | 3 Ph.D. in poultry science; 1 engineering technology grad
Team
4 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Research-to-commercialization; hardware prototype and AI models existed.

Stage 2: 5 formal submissions across 5 dates. These counts are descriptive, not scores.

Health / biotech

SpeechRoot

Formerly VoiceVault

Personalized synthetic voices for people facing speech or voice loss.

Learning in Stage 2: Clinical feedback, reproducible benchmarking and a pilot commitment informed the next steps.

University
Texas A&M University
College / school
College of Engineering
Academic profile
Undergraduate | Computer science
Team
1 listed member
Venture starting point and submissions

Original snapshot at the start of the challenge: Working production stack with one active beta user.

Stage 2: 7 formal submissions across 4 dates. These counts are descriptive, not scores.

Industrial / infrastructure

ZERO1 IO

Private AI agents for manufacturing operations.

Learning in Stage 2: Customer observation narrowed the delivery model toward a repeatable workflow with paid validation.

University
University of Silicon Valley
College / school
Master of Business Innovation program (college not specified)
Academic profile
Graduate | Master of Business Innovation
Team
2 listed members
Venture starting point and submissions

Original snapshot at the start of the challenge: Working hackathon prototype; no paid customers; selecting a first niche.

Stage 2: 5 formal submissions across 2 dates. These counts are descriptive, not scores.

Universities represented

InstitutionFinalist teams with this affiliation
Texas A&M University5
The University of Tulsa1
Tulane University1
University of California, Berkeley1
University of Wisconsin-Madison1
Baylor College of Medicine1
Columbia University1
University of Texas at Austin1
High Point University1
University of Silicon Valley1

Do Si represents Tulsa and Tulane; NEXUS represents Baylor College of Medicine and Columbia. Affiliations total 14 team-institution links across 10 distinct institutions, not 14 teams.

Affiliations, academic profiles and team sizes follow the corrected public finalist roster. Short learning summaries are based on the approved finalist descriptions. Read the finalist announcement.

THE PEOPLE BEHIND THE QUESTIONS

Judgment that moves ventures forward.

Our judges bring their experience to the evidence: asking what matters, testing the reasoning and helping founders see what deserves their attention next.

The judging community

Thank you to the entrepreneurs, investors, AI practitioners, educators and students who have contributed their time and judgment throughout the challenge.

Sarah Elk

Sarah Elk

Partner and Americas AI Practice Leader | Bain & Company

Sarah leads Bain’s AI Practice in the Americas and previously led its global People and Organization practice. She advises companies navigating technology and business-model disruption, with experience in agile innovation, organizational effectiveness and change management. Her perspective connects the potential of AI with the human decisions and operating changes needed to create real value.

Education: MBA, Stanford Graduate School of Business; BS, Environmental Engineering, Northwestern University

Professional profile
Jeff Bussgang

Jeff Bussgang

Co-founder and General Partner, Flybridge | Harvard Business School

Jeff is an entrepreneur turned venture capitalist and a co-founder of Flybridge. He teaches entrepreneurship and venture capital at Harvard Business School and is the author of The Experimentation Machine, Mastering the VC Game and Entering StartUpLand. His experience spans founding companies, early-stage investing and helping entrepreneurs build in large markets.

Education: MBA, Harvard Business School; BA, Computer Science, Harvard University

Professional profile
Ash Maurya

Ash Maurya

Creator of Lean Canvas | LEANSpark

Ash created Lean Canvas and wrote Running Lean. His work helps founders identify risky assumptions, test ideas with customers and validate demand before investing heavily in building. He now focuses on combining AI with disciplined startup validation to help founders turn domain expertise into viable ventures.

Education: BS, Electrical Engineering, Rochester Institute of Technology

Professional profile
Mitra Miller

Mitra Miller

Venture investor | Houston Angel Network

Mitra has more than 20 years of experience backing startups in disruptive technology, cleantech, new materials and life sciences. Her background includes operational, executive and advisory roles, alongside mentoring founders and investors. She also founded Eagle Investors, which connects students from Title I high schools with the startup community.

Education: MBA, Strategic Management, University of North Texas; BA, Rice University

Professional profile
Julian Bharadwaj

Julian Bharadwaj

AI systems and data science | Google

Julian brings more than two decades of experience in data science, machine learning platforms and large-scale software systems. His career includes Google, PayPal and Sabre, with work spanning autonomous agents, experimentation, causal inference and model governance. His engineering background informs a first-principles approach to evaluating technical systems and the evidence behind them.

Education: MEng, Industrial Engineering, Texas A&M University; BEng, Mechanical Engineering, Annamalai University

Professional profile
Ujjwal Rajbhandari

Ujjwal Rajbhandari

VP, Solutions & Delivery | Autonomize AI

Ujjwal builds and scales engineering teams that take enterprise AI from pilot to production. At Autonomize AI, his work focuses on deploying agentic AI in healthcare workflows. Previously a co-founder and CTO of Qubrid AI, with leadership experience at Google and Dell, he brings expertise in AI infrastructure, delivery and commercial execution.

Education: MS, Electrical Engineering, Texas A&M University; BE, Electrical Engineering, IIT Roorkee

Professional profile
Sammy Tao

Sammy Tao

Product management executive | AliveCor

Sammy brings experience planning, developing and launching products across consumer, mobile and automotive markets. His background includes NVIDIA, Amazon and Marvell Semiconductor. He works at the intersection of customers, engineering and partners, with expertise in product strategy, go-to-market planning and translating customer needs into differentiated products.

Education: MS, Electrical and Computer Engineering, Carnegie Mellon University; BS, Electrical Engineering, Texas A&M University

Professional profile
Jason (Zixuan) Jia

Jason (Zixuan) Jia

Software engineering | Netflix; previously LinkedIn

Jason is a software engineer with experience building AI-powered systems for marketing and advertising operations. His work includes agentic systems that combine language models with structured business logic to automate workflows, identify issues and improve data integrity. He brings a practical perspective on moving AI from a promising capability into useful software.

Education: BS, Computer Science, Texas A&M University

Professional profile
Yash Patel

Yash Patel

AI Engineer | Ciroos; previously Meta

Yash is an AI engineer and Texas A&M computer science graduate. He is interested in how AI models work under the hood and in exploring new technical approaches. His academic background spans machine learning, distributed systems and cloud computing, bringing a hands-on technical perspective to the judging community.

Education: MS, Computer Science, Texas A&M University; BTech, Computer Science and Engineering, Vellore Institute of Technology

Professional profile
Sai Nagabhairava

Sai Nagabhairava

Staff Software Engineer | Meta

Sai has more than 15 years of experience building large-scale distributed systems and infrastructure for cloud and AI workloads. His career includes Meta, Workday, PIMCO and Goldman Sachs. His work spans AI-agent safeguards, privacy, reliability and technical leadership, bringing a systems perspective to responsible and dependable AI deployment.

Education: BTech, Indian Institute of Technology Roorkee

Professional profile
Betty Adams

Betty Adams

Investor and finance educator | Mays Business School

Betty brings an investor’s perspective and experience teaching finance at Mays Business School. Her background in accounting, marketing and public administration offers a foundation for examining business models, financial assumptions and the evidence behind a venture’s plans.

Education: MPA, Columbia University; BA, Accounting and Marketing, University of Washington

Professional profile
Ben Wiggins

Ben Wiggins

Entrepreneur and investor | Calumet

Ben brings experience as a founder, angel investor and mentor. His company Proclaim was acquired by Sellerant in 2025. He also manages family-office assets and launched the Founders’ Group to connect entrepreneurs with mentorship and investors. His perspective combines company building, customer growth and early-stage investment.

Education: MBA and MS, Texas A&M University; BA, Economics, Rice University

Professional profile
Tyler Wooten

Tyler Wooten

Founder | FocusAI

Tyler is the founder of FocusAI and a Texas A&M mechanical engineering graduate. He is motivated by building things that can change the world for the better. He brings a founder’s perspective and an engineering background to conversations about how student ventures can turn promising ideas into meaningful progress.

Education: BS, Mechanical Engineering, Texas A&M University

Professional profile
Jack Roehr

Jack Roehr

Associate | Houston Angel Network

Jack brings a background spanning computer science, economics and finance, including venture capital, private equity and quantitative finance. His interest is in bringing innovative technologies to market. That combination offers both a technical and commercial perspective on the assumptions student founders need to validate.

Education: BS, Computer Science and Economics, Texas A&M University

Professional profile
Elizabeth Haegelin

Elizabeth Haegelin

Early-stage investing | Texas A&M graduate

Elizabeth focuses on early-stage investing, emerging technology and the founders behind new ventures. Her interests include research, diligence and helping founders move faster. At Texas A&M, she studied Business Honors and Finance with a minor in Entrepreneurship and participated in the Aggie Venture Fund and other investment programs.

Education: Business Honors and Finance; minor in Entrepreneurship, Texas A&M University

Professional profile
Mahadev Annabhimoju

Mahadev Annabhimoju

Economics student | Vanderbilt University

Mahadev joined us as a freshman undergraduate business student at Mays Business School with a strong interest in venture capital. We promised him real-world experience and learning if he leaned in and helped with AIVVC. He did exactly that: volunteering his time and becoming instrumental in the successful launch of the AI Venture Velocity Challenge. Now studying economics at Vanderbilt, he remains part of our judging community. We believe he will be an amazing venture capitalist, or excel in whatever he chooses to pursue!

Education: Economics, Vanderbilt University; previously Texas A&M University

Professional profile

Evidence first. Human judgment throughout.

Judges evaluate problem significance, learning velocity, AI leverage, venture progress, responsible design and adaptability. In the finals, their questions help clarify what the evidence demonstrates and what the founders should test next.

Earlier rankings do not carry points into the prize decision. The complete venture record remains part of the evaluation.

Recognition reflects contributions across the challenge, not a list of confirmed live-finals attendees. Affiliations are provided for context and do not imply employer endorsement.

* Naisha Kumar, Centennial High School, participated as a guest judge, representing the next generation of student entrepreneurs.

SEPTEMBER 26 / COLLEGE STATION

The evidence continues into the finals

All 12 finalists have an opportunity to win. Semifinal rankings determined advancement; they do not carry points into the prize decision.

A conversation, not a pitch contest

Judges will consider the complete venture record, the final update and a live discussion under the existing rubric. Founders will explain what matters most to prove next, respond to questions and reflect on what the conversation changes.

  1. 90 seconds: Venture opening
  2. 10 minutes: Conversation with judges
  3. 1 minute: Quiet reflection
  4. 90 seconds: Closing reflection
  5. 1 minute: Moderator wrap-up and transition

After the finals

A fuller learning report can examine the quality of experiments, how customer evidence changed decisions, the role of human judgment and how ventures progress beyond the event. Long-term outcomes and stronger claims about AI-enabled acceleration require further evidence.

Meet the judging community

The 2026 AI Venture Velocity Challenge is supported by Deloitte Foundation, Deloitte, and Midjourney.

Updated September 15, 2026. Earlier chapters retain their original Stage 1 scope.