Kindo Raises $30 Million to Tame Enterprise AI Agent Sprawl

Kindo has raised $30 million to expand its enterprise AI agent management platform, bringing total funding to $57.6 million — money wagered on the assumption that most employers deploying AI agents (software programs that act autonomously against business systems) have little idea what those agents can access, do, or spend.

The round was co-led by Crosspoint Capital and Forgepoint Capital, with Andre Fuetsch and Ernie Bio, managing directors at the two firms respectively, joining the Venice Beach, California company's board. Kindo reports roughly 800% year-over-year revenue growth in 2026 — a self-reported figure, and one that reflects a small base growing fast more than an established market. TFX Capital and CircleRock Capital participated, alongside existing investors including Drive Capital, Riot Ventures, RRE Ventures, Eniac Ventures, Marlinspike Partners, New Era Ventures and Sunset Venture.

What problem is the money buying into?

Deploying an agent delegates authority. An agent connected to business systems may read internal information, update records, approve requests, or modify infrastructure. As deployments multiply, security and operations teams need visibility into both the permissions an agent holds and the actions it actually takes — including the ones nobody sanctioned.

Founder and CEO Ron Williams said the gap is not a shortage of agent builders. It is the tooling organizations need to control how agents are built, which data they can reach, where information can go, and what operating those agents costs.

Kindo positions its platform as a shared operating layer for IT, security, finance and compliance — an "AI Operations Center" that applies the continuous monitoring model of security operations centers to enterprise agents. Discovery maps sanctioned and shadow AI (tools adopted without approval), while a central view consolidates activity, access, costs and outcomes. A registry tracks agent ownership, permissions inherited from the person running an agent, and cost attribution. Teams can build agents inside Kindo or connect agents built elsewhere.

How deep does the control actually go?

Two tiers, and the difference matters. Gateways govern and audit agent calls to connected data, tools and approved models, applying existing organizational permissions to both people and agents. That makes each connection a policy control point rather than leaving every agent team to build its own disconnected controls.

For mission-critical deployments, Kindo offers a purpose-built agent harness — the execution environment that manages an agent's run and lets operators monitor, support, intervene, and take over an agent when something goes wrong. The release specifically ties deeper intervention and several cost features to agents running inside the harness. Connect an external agent through gateways and you get governance; run it in the harness and you get control. Customers can deploy on Kindo's cloud, a self-managed cloud, on-premises, or air-gapped.

Can finance finally see the bill?

The AI Control Center records token consumption and agent-run outcomes, including failed runs, attributing spend to individuals, agents, departments and projects. Finance teams can set budgets, receive alerts and enforce hard caps. The accounting matters because sticker price is a poor proxy for value: a cheap agent that repeatedly fails wastes money, while an expensive model may be justified for a difficult task.

The Deep Hat family targets four distinct cost sources:

• Sift learns which events an agent would dismiss, then filters them after a person approves the threshold — starting in shadow mode and sending a sample of filtered events through to catch misses. Its accuracy is measured against the target agent's verdicts, not independently verified ground truth.
• Tune adapts a smaller, cheaper model to a specific task, reserving large models for work that needs them.
• Shrink runs repeatable workflow steps as code, removing model cost from those steps and routing exceptions back to a model.
• Attack supports private, authorized offensive security testing instead of using frontier cybersecurity models for routine work.

Who is actually using it?

Kindo says the platform runs in production at large and mid-market enterprises across financial services, professional services, healthcare, and aviation and transportation. The announcement highlights Aireon, whose CISO reported in May 2026 that using Kindo for threat hunting and Tier 1 and Tier 2 security operations helped generate more than $2 million in annual savings — a customer-reported outcome, not a guaranteed result elsewhere.

Kindo's funding round rests on a durable asymmetry: building agents is getting easier, while supporting them in production still requires ownership, access policies, intervention mechanisms and spending controls. Whether $30 million is enough to make those responsibilities manageable across agents built in different places is the question its go-to-market expansion now has to answer.