thegeekconf mini 2026
The thegeekconf mini 2026 showcased talks on creating production-grade AI experiences, transitioning from prompt prototyping to robust software engineering, and exploring agentic architectures. Experts demonstrated building intelligent features with AWS and Microsoft tech stacks and introduced framework paradigms for secure model governance and token control.
SNAPSHOT
Keynote: Architecture Over Models
Suresh Konakanchi explored the realities of shipping AI in production in his keynote at thegeekconf mini 2026. The talk highlighted a critical truth in product engineering: AI products rarely fail because of underlying LLM models, but rather due to systemic brittle points like poor context management, hidden token costs impacting P&L, and silent model updates breaking prompts. Its focus on identifying these hidden operational risks and security gaps underscores the absolute necessity of designing AI-first systems from the ground up to survive enterprise scale.
Key Takeaways:
Driven by engineering discipline and tools like AI gateways and purpose-built Small Language Models (SLMs), AI product engineering continues to push boundaries into areas like automated agentic workflows and multi-modal systems. While challenges like observability and prompt injection persist, the ecosystem's maturity ensures ongoing progress. The keynote reinforced the vision of moving past surface-level AI hype, empowering AI developers to design scalable, production-ready architectures seamlessly.