The Stack Overflow Podcast After all the hype, was 2025 really the year of AI agents?
10 snips
Mar 20, 2026 Stefan Weitz, CEO and HumanX co-founder focused on AI and developer platforms. He debates whether 2025 delivered on AI agents and why AGI hype cooled. He highlights infrastructure gaps, trust and data readiness blocking adoption. He also explores how models shift human-computer interaction and the practical limits of vibe-coding and agent architectures.
AI Snips
Chapters
Transcript
Episode notes
Agents Fell Short Of The Utopian Promise
- Agents were widely hyped as 2025's breakthrough but failed to deliver a universal utopia.
- Stefan Weitz says 2025 shifted discussion from hype to practical limits like multi-node management, infra and runtime constraints.
Three Concrete Barriers To Agent Adoption
- Three core gaps block agent adoption: infrastructure, trust, and machine-readable data.
- Weitz details missing AI-ready datacenters, multi-node agent architectures, trust infrastructure, and legacy enterprise data silos.
Legacy Data Prevents Agentic Access
- Enterprise data rarely exists in machine-readable form suitable for agents.
- Weitz cites ETL, batch jobs, flat files and AS400 systems needing transformation before agents can consume them reliably.