Many AI pilots succeed in limited tests but stall because the underlying technology lacks the enterprise-grade scale, rigor, and compliance required for full production. Moving from 10 to 10 million interactions is a fundamentally different challenge that trips up many programs.

The original playbook of simply scaling parameters and data is now obsolete. Top AI labs have pivoted to heavily designed post-training pipelines, retrieval, tool use, and agent training, acknowledging that raw scaling is insufficient to solve real-world problems.

Resist building complex, multi-agent systems from day one. Instead, start with a single agent and build its skills based on actual workflows. Add sub-agents only when a clear productivity need arises. This approach is more effective than scaling for what looks impressive.

With AI agents capable of generating code and designs at an unprecedented rate, the new chokepoint in workflows is human review. The primary challenge is no longer production but scaling the evaluation process to ensure AI-generated output aligns with quality standards and company values.

AI product quality is highly dependent on infrastructure reliability, which is less stable than traditional cloud services. Jared Palmer's team at Vercel monitored key metrics like 'error-free sessions' in near real-time. This intense, data-driven approach is crucial for building a reliable agentic product, as inference providers frequently drop requests.

The 'move fast and break things' mantra is counterproductive for complex AI development. Tools and philosophies prioritizing correctness and thoughtful architecture over raw speed are better suited for building meaningful, non-trivial AI features that don't become overwhelming to manage.

AI agents bypass traditional user onboarding, immediately push software to its limits, and even request missing API functionality. This requires a fundamental shift in product development, focusing on robust, agent-friendly infrastructure rather than just human-centric UIs.

Many developers believe tweaking prompts and logic ('harness engineering') is the hardest part of building agents. The real bottleneck, however, is scaling, reliability, and managing production infrastructure—a common miscalculation that managed services aim to solve.

While AI proofs-of-concept are easy, SAP's CTO states the real engineering hurdle is scaling reliably. The complexity lies in managing thousands of APIs, handling massive document volumes, and applying granular, user-specific context (like regional policies) consistently and accurately.