Episode 16: Building AI-Driven Futures

Video not loading? Try accepting cookies in the bottom right corner.

Executive Summary

In this episode of Databytes Enabling AI – Prowess Consulting, recorded in collaboration with Venture Studio, Ahmedabad University, host Rahul Nawab, Jeremy Fritzhand sits down with Hemen Ashodia, Founder and CEO of FXIS AI, for a deep and refreshingly honest conversation on what it really takes to build AI that works. A pioneer who entered the field in 2012 — long before AI became a boardroom buzzword — Hemen traces his journey from teaching professors about neural networks to leading transformative AI projects for global enterprises including Johnson & Johnson, AB InBev, and Bitcoin.com. He challenges the widespread habit of brute-force model switching, instead making a compelling case for single-variable experimentation, first-principles thinking, and understanding the mathematics beneath every model. Through vivid examples — from a legal case simulation tool to a psychotherapy AI evaluation framework — Heman illustrates the iterative, sculpting-like nature of real AI engineering. He also addresses the hard limits of current AI with numbers and visual data, why coding is far from fully automated, and what it means to build a true moat as an AI-first founder.

About Hemen Ashodia

Hemen is the Founder & CEO of fxis.ai, where he has led transformative projects across generative AI, explainable AI, and automation for global giants including Johnson & Johnson, Bitcoin.com, Tala, and AB InBev. His technical acumen and innovation mindset have positioned him at the forefront of AI adoption—particularly in designing optimized data retrieval systems for cloud efficiency and enterprise scalability. He is also the Co-founder of Remarkin.com, a globally used, gamified e-learning platform, and a Venture Design Fellow through a program aligned with Stanford University, where he has mentored the next generation of AI thinkers and innovators.

What You'll Learn

  • Why mathematics — not coding or model downloads — is the true foundation of AI engineering, and what separates real engineers from those doing “data black magic”
  • How to approach AI development as an iterative, sculpting process — and why chasing perfection too early kills most projects
  • A practical single-variable experimentation framework for choosing the right model size, company, and reasoning approach without brute-force guessing
  • The real limitations of current AI — from numerical reasoning and visual data to the quadratic attention problem that caps context window performance
  • What it means to build a moat as an AI founder — and why injecting a chatbot is not a business strategy
  • Career advice for non-technical students: how to use tools like Lovable, Replit, and Claude Code to build, ship fast, and validate ideas before hiring a tech team

Who Should Watch

  • Industry leaders looking for strategic insights in data, AI, and marketing to guide high-level initiatives and innovation.
  • Working professionals seeking practical, actionable knowledge to build expertise and stay ahead in their fields.
  • General audience interested in AI trends and thought leadership, including early-career professionals and curious learners.

Databytes Enabling AI: Series Overview

Stay tuned for more episodes!

Related Posts

Never miss a story

Subscribe to the blog and stay updated on Prowess news as it happens.