How Anyone Can Build Meaningful AI Without Code - Ep. 283
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
In this episode, Shanea Leven, CEO and co-founder of Empromptu AI, shares how her company is revolutionizing AI development by making it accessible to non-technical users while achieving up to 98% accuracy in AI outputs. (02:49) Shanea discusses her journey from Google and Docker to building a platform that lets anyone create production-ready AI applications without extensive coding knowledge. The conversation explores how Empromptu's "AI that builds AI" approach is transforming businesses across industries, from CPG brands using ocean plastic to create activewear to mom-and-daughter teams building financial literacy apps. (05:24) The episode delves deep into enterprise AI transformation, the importance of provable AI for building trust, and how NVIDIA CUDA powers their optimization engine to deliver real-time, accurate results for millions of users.
- Core themes include democratizing AI development, bridging the gap between technical and non-technical users, achieving enterprise-scale AI accuracy, and transforming businesses through accessible generative AI tools.
Speakers
Shanea Leven
Shanea Leven is the co-founder and CEO of Empromptu AI, with an extensive background spanning Google, eBay, Cloudflare, and Docker. She studied business and computer science, then spent time at Google working on developer tools for Google Home and Android, helping build applications for millions of developers worldwide. (01:04) Previously, she founded CodeSee, which was acquired, and served as head of product at multiple companies before co-founding Empromptu with computational physicist Dr. Sean Robinson.
Noah Kravitz
Noah Kravitz is the host of the NVIDIA AI podcast, guiding conversations about cutting-edge AI developments and their real-world applications across various industries.
Key Takeaways
Transform Legacy Systems with AI Integration
Rather than starting from scratch, successful AI transformation involves integrating AI capabilities into existing business infrastructure. (07:00) Shanea explains that most enterprise clients already have SaaS applications, platforms, or running businesses, and the key challenge is "how do I transform that into an AI native company?" This approach respects existing investments while enabling dramatic capability upgrades. The practical application involves using specialized tools that can ingest entire code bases from GitHub and add AI functionality without requiring complete rebuilds.
Focus on Task Success Over Model Benchmarks
Empromptu redefines AI accuracy by measuring "task success" rather than relying solely on model benchmarks. (23:59) This means allowing users to define what success means for their specific use case, then optimizing the entire system—including models, data, prompts, and evaluations—toward that user-defined goal. This approach recognizes that enterprise users care more about whether AI helps them accomplish their business objectives than abstract performance metrics.
Embrace Mixed-Code Development Philosophy
The future of AI development lies in "mixed-code" approaches that combine technical and non-technical capabilities rather than forcing a choice between no-code and pro-code solutions. (20:37) This philosophy acknowledges that both technical veterans with 20 years of experience and newcomers are learning generative AI simultaneously, creating opportunities for collaborative development where AI handles complex technical tasks while humans focus on creative problem-solving and strategic direction.
Implement Provable AI for Trust Building
Building trust in AI systems requires transparency through "provable AI" approaches that show users exactly what decisions the AI is making and how. (26:20) This includes providing visibility into data sources, decision-making processes, rollback capabilities, and accuracy improvements over time. Users need to see, feel, and understand AI behavior in concrete terms, with dashboards showing real numbers and suggestions rather than black-box outputs.
Leverage Computer Science Fundamentals for AI Success
Despite headlines suggesting coding skills are becoming obsolete, computer science education remains crucial for AI development because it teaches critical thinking, systems thinking, and problem decomposition skills. (33:01) Shanea emphasizes that computer science "is about cognition, it's about thinking in systems, it's about critical thinking skills, it's about how do you break problems down into small shippable chunks." These foundational skills become even more important when conducting AI orchestras rather than writing every line of code manually.
Statistics & Facts
- Empromptu can achieve up to 98% accuracy in AI outputs using their optimization technology. (03:03) This statistic was mentioned when Shanea first met her co-founder Dr. Sean Robinson, who claimed he invented technology to reach this accuracy level, leading to their partnership and company formation.
- The company can deliver production-ready chatbots in approximately 10 minutes using their self-serve platform. (14:13) This demonstrates the dramatic reduction in development time compared to traditional approaches while maintaining enterprise-scale functionality.
- After about 30 manual optimization runs, Empromptu can switch to automatic optimization mode to reach their 98% task accuracy target. (25:04) This provides a clear pathway for less technical users to achieve sophisticated results without deep technical knowledge.
