AI for FinOps: From Hype to Helpful, FinOps X 2026 Sneak Peek

Why It Matters

Because AI is becoming a core FinOps capability, disciplined use can drive cost efficiency while preventing costly errors from unchecked model outputs.

Key Takeaways

• Provide detailed context to LLMs for accurate FinOps queries.
• Use constrained, specific prompts to avoid hallucinations in cost analysis.
• Treat AI as an abstraction layer; verify outputs manually.
• Build a personal AI “home base” with persistent sessions and custom skills.
• Leverage AI to automate billing data extraction, but maintain skeptical oversight.

Summary

The video is a sneak‑peek of a forthcoming FinOps X presentation titled “AI for FinOps: From Hype to Helpful.” The speaker frames AI as a new baseline technology that finance‑operations teams must master to stay competitive, and promises actionable guidance beyond the high‑level hype.

He stresses three practical pillars: supplying rich context (schema, labels, billing export), issuing tightly scoped prompts, and maintaining a skeptical stance by double‑checking results. Without these, large language models can hallucinate or misinterpret cost data, turning a powerful tool into a liability.

Real‑world examples include copying a generated BigQuery query from the GCP billing console into an LLM to refine it, and constructing a personal “AI home base” using tmux, Claude Code, and custom skill directories. This setup stores common queries, references, and memory, enabling a tailored assistant that persists across sessions.

For FinOps professionals, adopting these practices can automate routine cost‑analysis tasks, accelerate decision‑making, and free analysts for strategic work. However, the black‑box nature of LLMs demands rigorous validation, making disciplined prompt engineering and personal AI infrastructure essential for reliable outcomes.

Original Description

Dann Berg from Squarespace previews his FinOps X 2026 (https://x.finops.org) talk on bridging the gap between believing AI will change everything and knowing what to actually do on Monday morning. He covers why context is everything when using LLMs for FinOps work (schema, label taxonomy, billing export structure), why being a skeptic matters, and a framework for understanding AI as the first technology abstraction layer where trust is required but visibility is gone.

This is a teaser for his full session at FinOps X 2026 in San Diego.

Recorded at the April 2026 FinOps Foundation Virtual Summit.

WATCH the full Summit: https://www.youtube.com/watch?v=ZDWElwO0KSI

See the FinOps X 2026 Agenda: https://x.finops.org/agenda

Join the Community: https://finops.org/join

#aiforfinops #aioptimization #finopsx