VENTURE
DEVILS

Spring 2026 — Building Turing at ASU's Flagship Incubator

Venture Devils is ASU's entrepreneurship incubator, providing student-led startups with mentorship, funding access, and structured business development support. During Spring 2026, I am developing Turing, an AI-powered desktop automation tool that observes user workflows via screen recordings and learns to replicate them autonomously. The program involves weekly mentorship sessions, business model development, customer discovery interviews, and pitch preparation.

Problem Statement: Knowledge workers spend significant time on repetitive computer tasks, yet existing automation tools require technical API configurations that most users cannot implement. Turing addresses this gap by using Vision Language Models to extract workflow patterns from screen recordings, enabling task replay without any API configuration.

Prior to Venture Devils, my experience was limited to building technical solutions without systematic validation of user needs. The customer discovery process introduced a structured methodology for evaluating product-market fit: conducting interviews, analyzing feedback patterns, and iterating on the business model accordingly. This entrepreneurial skill set complements my research and coursework training.

Venture Devils has reinforced the interdisciplinary nature of my GCSP experience. Research methodology from FURI informed Turing's algorithm design. The PESTLE framework from FSE 150 and the technology-society analysis from STS 330 shaped my consideration of how automation tools affect different worker populations. Entrepreneurship through GCSP is fundamentally about learning to build solutions that are sustainable and responsive to genuine human needs.

VALUE

RELATION
TO MY
THEME

Human error causes 95% of cybersecurity breaches. A lot of that is password fatigue and people taking shortcuts because they're tired of typing the same credentials over and over. Turing tackles this by automating those repetitive tasks with on-device processing, so sensitive data never leaves the user's machine. This connects to my earlier work on Flashback at HackMIT, where I built privacy-first facial recognition. Both projects share the same principle: you can make things more convenient without sacrificing Security. Turing's architecture is built around strict data localization because I don't think productivity tools should come at the cost of user trust.

contact@aryankeluskar.com
aryankeluskar.com