Stacklet at KubeCon 2026: AI Cloud Governance Gets Serious

The News

At KubeCon North America 2026, Stacklet co-founder and CEO Travis Stanfield outlined a set of new product announcements centered on AI and cloud governance. The company introduced AI policy packs for managing cost and risk in AI workloads, a capability called Token Custodian that extends the Cloud Custodian open-source project to token and tokenomic use cases, and a coding agent policies and skills feature that surfaces Stacklet’s governance knowledge base directly to AI coding agents such as Claude. The announcements reflect Stacklet’s broader positioning as a unified governance layer that spans cloud infrastructure, Kubernetes clusters, and AI workloads across multi-cloud environments.

Analyst Take

The governance gap that AI is making worse

Stacklet is addressing a problem that didn’t exist at this scale two years ago: AI agents are now the primary deployers of cloud infrastructure. When Stanfield says “AI is the primary deployer of cloud these days,” he’s describing a structural shift in how infrastructure gets created, and governance frameworks haven’t kept pace. Infrastructure-as-code generated by AI agents carries the same misconfiguration risks as human-authored code, but at a volume and speed that manual review processes simply can’t match. This isn’t a niche concern for platform engineers. It’s a CFO and board-level conversation about whether the organization can actually control what it’s spending and what attack surface it’s accumulating.

The multi-cloud reality compounds the challenge. Most enterprises aren’t running a single cloud. They’re running several, often with different teams, different toolchains, and no shared governance vocabulary. Stacklet’s Cloud Custodian heritage gives it a defensible position here: a policy-as-code foundation that can operate consistently across AWS, Azure, GCP, and the data cloud providers (Snowflake, Databricks) that are increasingly part of the AI infrastructure stack.

Token economics as the new FinOps frontier

The Token Custodian announcement is the most strategically interesting of the three. FinOps matured as a discipline because cloud spend became opaque and wasteful without it. Token spend is following the same curve, faster. Organizations that have broadly deployed AI copilots and agents are already discovering that not every inference needs to run on the most expensive frontier model. Stanfield’s framing of “model routing” is the right abstraction: applying business context to route workloads to appropriately priced models is exactly the kind of optimization that separates AI programs that deliver ROI from those that just generate invoices.

This connects directly to what ECI Research found in its Google GovTech Survey: according to ECI Research, 49.6% of respondents estimated that 26% to 50% of their organization’s code would be AI-assisted or generated within the next 12 months. As AI-assisted development scales to that level, the volume of inference calls across the software delivery lifecycle becomes a material cost line. Organizations that lack per-workload attribution and dynamic budgeting controls will face the same sprawl problem that caught cloud teams off guard in 2015–2018.

The coding agent integration is a distribution play

The coding agent policies and skills feature deserves attention from developers specifically. Surfacing governance context directly inside tools like Claude-powered coding agents means that policy enforcement moves from a gate at the end of a pipeline to a guardrail at the point of creation. That’s the right direction. ECI Research’s Google GovTech Survey found that 31.0% of respondents reported that security is treated as a final gate, causing significant rework. Shifting governance knowledge into the agent’s context window is one credible path to breaking that pattern, because it changes the agent’s default behavior rather than relying on developers to consult documentation or remember policy requirements mid-sprint.

The coding agent integration also functions as a distribution strategy. Developers who are already using AI coding tools are not going to open a separate governance console to check compliance. Meeting them inside the tools they’re already using is where adoption happens. Stacklet’s ability to inject its curated policy library, including customer-customized policies, into that context is a meaningful differentiator over point solutions that only offer generic rulesets.

For ITDMs, the financial framing Stanfield offered is worth taking seriously. He explicitly positioned the product as solving the tension between “we want all the value of AI” and “we don’t want the organization to go broke.” That’s an accurate description of the budget conversation happening in most executive teams right now, and governance tooling that can quantify and constrain AI infrastructure spend has a clear ROI story.

Looking Ahead

Stacklet’s roadmap toward autonomous governance, where agents detect upstream changes from cloud and model providers and automatically infer, test, and deploy the appropriate guardrails, is the right long-term direction but will require organizational trust to execute. The harder problem isn’t the technology; it’s convincing security and compliance teams to let an automated system make governance decisions without a human sign-off. The customers Stanfield describes as already “automated” and now moving toward “autonomous” are likely large enterprises with mature DevSecOps practices. Broader market adoption will depend on Stacklet building explainability and audit trails into autonomous decisions that can satisfy both internal risk teams and external auditors.

The competitive landscape will get crowded. Hyperscalers are not going to cede the governance layer on their own platforms without a fight, and security vendors are extending toward cloud posture management from the other direction. Stacklet’s durable advantage is the Cloud Custodian ecosystem, its multi-cloud and multi-paradigm reach from pod to pipeline, and the institutional knowledge embedded in its policy library. If it can translate that into a compelling SaaS offering that enterprises can adopt without a lengthy professional services engagement, it has a real opportunity to own the governance category as AI infrastructure spend becomes one of the largest and least-controlled line items on the modern enterprise balance sheet.

Authors

  • Paul Nashawaty

    Paul Nashawaty, Practice Leader and Lead Principal Analyst, specializes in application modernization across build, release and operations. With a wealth of expertise in digital transformation initiatives spanning front-end and back-end systems, he also possesses comprehensive knowledge of the underlying infrastructure ecosystem crucial for supporting modernization endeavors. With over 25 years of experience, Paul has a proven track record in implementing effective go-to-market strategies, including the identification of new market channels, the growth and cultivation of partner ecosystems, and the successful execution of strategic plans resulting in positive business outcomes for his clients.

    View all posts
  • Samantha Weston

    With over 15 years of hands-on experience in operations roles across legal, financial, and technology sectors, Sam Weston brings deep expertise in the systems that power modern enterprises such as ERP, CRM, HCM, CX, and beyond. Her career has spanned the full spectrum of enterprise applications, from optimizing business processes and managing platforms to leading digital transformation initiatives.

    Sam has transitioned her expertise into the analyst arena, focusing on enterprise applications and the evolving role they play in business productivity and transformation. She provides independent insights that bridge technology capabilities with business outcomes, helping organizations and vendors alike navigate a changing enterprise software landscape.

    View all posts