A year ago, Palo Alto Networks built dashboards to track AI spend. Today those dashboards are useless, and the team that built them thinks that's the whole story. Recorded at FinOps X in San Diego, this conversation brings together Abhinav Lad, who leads cloud and AI finance at Palo Alto Networks, and Kuntal Patel, who runs the cloud engineering function behind it. They explain what happened when agents entered the picture, and AI stopped behaving like a service anyone could forecast.

This blog sketches the AI stack as it’s emerging from real‑world practice and invites the community to refine it. The goal is to create a shared, peer‑validated blueprint for anyone building AI systems, shaped by real experiences, not marketing claims.

When enterprises build AI agent platforms, selecting the right gateway is critical for routing models, models tools (like MCP servers), and securing traffic. While LiteLLM is widely recognized for its extensive provider catalog and unified OpenAI-compatible API, it often falls short in enterprise evaluations because crucial security and platform controls—such as SSO beyond five users, secret manager integrations, scoped guardrails, and audit logging—are locked behind a proprietary enterprise license. Conversely, agentgateway (an Apache 2.0 open-source project hosted by the Linux Foundation) serves as a robust, general-purpose data plane for HTTP, gRPC, and LLM traffic, providing enterprise-grade security features without embedded license gates. Furthermore, agentgateway boasts a stronger public security record, avoiding the critical vulnerabilities that have plagued LiteLLM. For teams that need a true microservice gateway for AI services rather than just an expansive SDK, agentgateway presents a more secure, open, and scalable architectural choice.

In this episode of Agentic Conversations, we sit down with Ambud, Principal Engineer at Pinterest responsible for general technology efficiency, fresh off delivering a controversial keynote on AI infrastructure optimization at scale. Ambud walks us through his Five Layer Cake framework - a structured approach to driving efficiency across every level of the AI stack, from silicon and hardware procurement to model selection, inference engine design, and governance. We explore how decisions compound across layers to unlock real business growth, and how the wrong choices can lock you into expensive commitments for years.

Preparing an MLOps repository for AI coding agents requires eliminating friction points that cause AI hallucinations, such as fragmented toolchains and conflicting CI/local scripts. To create an "agent-ready" environment, teams should adopt a "single-gate pattern" that unifies execution commands across all environments, while ensuring that standard AGENTS.md instructions are strictly enforced by automated, machine-readable rules. Additionally, optimizing test performance for isolated MLflow environments—such as using a SQLite template database pattern—and managing dependency vulnerabilities through explicit, tested overrides are crucial. Ultimately, building a deterministic, rigorously checked repository allows AI assistants to bypass tooling drift and efficiently produce correct, production-ready code.

An analysis of the upcoming 2026-07-28 MCP specification updates. It covers how the protocol is moving from stateful handshakes to a stateless core, allowing remote servers to run behind standard round-robin load balancers. The piece also details the new enterprise-managed token exchange (ID-JAG via RFC 8693) and the release of beta SDKs with structured 12-month deprecation cycles.

Every engineer at Wayfair can now see, in real time, exactly what their code costs, and that's on purpose. Brent Eubanks, FinOps Architect at Wayfair, walks us through what happens when you stop treating AI spend as a finance problem and start treating it as an engineering one. The story that sticks with you: a team was burning $400k a month on an LLM-driven workflow, until they flipped the whole thing on its head - hard-coded logic doing the heavy lifting, AI called in only when it's actually needed. New spend: $80k. It's the kind of move that only makes sense once you realize more agentic autonomy isn't always the win it's sold as. From there we get into the machinery Wayfair has built to make cost ownership feel less like a punishment and more like a superpower: guardrails baked straight into coding copilots, predictive alerts that catch a team blowing its budget before the month even ends, a "spend captain" system that pushes budget decisions down to the people closest to the work, and a genuinely strange new question every tech leader is starting to ask: if you were staffing a company with a team of AI agents instead of people, what would that org chart even look like?

AI models are starting to act like appliances, locked into one narrow way of working, instead of the flexible infrastructure they used to be. Drew Breunig, an AI and data strategist working with the Overture Maps Foundation, joins us to explain why, and what it means for anyone building something that doesn't look like Claude Code. Drew walks through his "Winchester Mystery House" idea: what happens once code gets so cheap to write that the only real bottleneck left is feedback. From there we dig into DSPy: signatures, the GEPA optimizer, and the brand-new Flex optimizer, which rewrites your code instead of just your prompt, complete with a real before-and-after on cost and accuracy. We also get into why so many AI-built apps and websites end up looking identical, the actual difference between an agent and a workflow, what Drew learned a year after shipping a code library with no code in it, and why he thinks the most valuable thing you can do right now is close the laptop and go talk to people.