What is an AI Agent Control Plane?

An Agent Control Plane is a centralised control layer that combines visibility, access rights, runtime monitoring and cost management for AI agents across an organisation — comparable to a security and governance hub for every active agent.

Why did multiple vendors launch products on the same day?

CrowdStrike deliberately tied its launch to its own conference, Fal.Con. The launches from Boomi, Orchestry, Ping Identity, Kong and LogicGate were independent of that event — their coincidence shows that the entire industry is reacting, separately, to the same problem: agent sprawl in production.

What does AIDR (AI Detection and Response) mean?

AIDR is a new security category coined by CrowdStrike, specifically designed to detect and respond to risks created by AI agents — analogous to established categories like EDR for endpoints.

Does this matter for Swiss companies without US operations?

Yes. Once AI agents work with business or customer data, questions of access rights, traceability and accountability apply regardless of location. For companies with EU business, EU AI Act classification adds a further layer of relevance.

What is agent sprawl?

Agent sprawl describes a state in which more AI agents are active within an organisation than IT and business teams can track — often including self-built shadow agents with no central record.

Where should a company start if it has no governance in place yet?

With an inventory: which agents already exist, who created them, and which systems and data they touch. Only after that can you meaningfully decide which control-plane capabilities you actually need.

AI Insights

September 1, 2026: The Turning Point for AI Agent Governance

Chris Jon Graf · AI Strategist & CEOPublished on 2 September 2026

In short

On September 1-2, 2026, CrowdStrike, Boomi, Orchestry, Ping Identity, Kong and LogicGate independently launched control infrastructure for AI agents within 24 hours. This is the coordinated response to agent sprawl: agent governance has shifted from optional to a production requirement — and it is now available off the shelf.

Six Vendors, One Day: What Happened on September 1, 2026

Within 24 hours — on September 1 and 2, 2026 — six major vendors independently launched control infrastructure for AI agents: CrowdStrike with Falcon Guardian, Boomi with its Agent Control Plane, Orchestry with AI & Agents for Microsoft 365, Ping Identity with Enterprise Personal Agent Access, Kong with AI Gateway 2.0, and LogicGate with Agentic GRC. The same day brought further announcements from Ema, CBTS and Nutanix. This is not coincidence — it is the moment an entire industry recognised, simultaneously, that AI agents in production need a control layer, and that this layer is now available to buy.

CrowdStrike Falcon Guardian defines a new security category, «AI Detection and Response (AIDR)», covering agent access controls, runtime detection and response, and an AI gateway.
Boomi Agent Control Plane is AI-native infrastructure that securely connects agents to core business systems — running in public cloud, VPC or on-premises, a direct answer to data sovereignty concerns.
Orchestry AI & Agents gives IT teams one place to see every agent running on a Microsoft 365 tenant, including risk scoring and the ability to delete agents tenant-wide.
Ping Identity Enterprise Personal Agent Access discovers personal agents — including shadow AI — and ties every session to the user and device behind it.
Kong AI Gateway 2.0 bundles governed agent tool access through MCP servers and adds dynamic, modality-aware cost management.
LogicGate Agentic GRC makes GRC agents generally available for third-party risk, enterprise risk, AI governance and business continuity management.

Agent Sprawl: The Problem Everyone Is Solving at Once

The common thread behind all six launches is agent sprawl: agents are being rolled out faster than organisations can track them. Analyses published on September 1 warn that many companies barely know which agents are active in their systems, what data they touch, or which permissions they hold. That gap — between rollout and oversight — is the real trigger behind this governance wave.

75% / 5%

Deloitte 2026: 75% of organisations expect widespread AI agent adoption, but only 5% consider their processes ready for it.

40%+

Gartner forecast: more than 40% of agentic AI projects will be cancelled by 2027 due to governance gaps.

These figures explain why governance is no longer optional. Running agents in production without knowing how many exist or what they are permitted to do means operating with exactly the kind of gap that, according to Gartner, kills projects before they deliver value.

What an Agent Control Plane Actually Covers

Despite different branding, all six products address the same five control areas — the new baseline for agents in production.

Visibility: which agents are running, and where — including self-built or downloaded shadow agents.
Access control: which systems, data and actions an agent may touch, and who is accountable for it.
Runtime monitoring: detecting misbehaviour or misuse while an agent is active, not after the fact.
Cost and gateway control: centralised management of tool access and the costs it generates.
Governance workflows: risk scoring, approval processes and audit trails, the same discipline GRC teams already apply elsewhere.

The Swiss Context: Not Just a US Enterprise Story

Agent adoption is no longer a future question in Switzerland either: Helvetia became the first listed Swiss insurer to launch a ChatGPT-based customer service, and Ringier is working under an OpenAI Enterprise partnership. In parallel with this governance wave, the EU classified ChatGPT under the Digital Services Act on September 2 and designated OpenAI's Astra as «high risk» under the EU AI Act. For Swiss companies with EU business, that is not an abstract regulatory footnote — it is a direct benchmark for their own agent governance.

Governance Is a Boardroom Question

Agent governance is no longer purely an IT matter. How much AI literacy and decision authority needs to sit at leadership level is explored in the Swiss KI-Podcast episode on autonomous business decisions.

What This Means for Swiss Mid-Market Firms Right Now

The good news: you no longer need to build this control layer yourself. The six launches on September 1 show that agent governance has become a software category you can buy, rather than something you spend years engineering internally. The real work lies in first establishing which agents are actually active in your organisation, what data they touch, and who is accountable for them — before adopting one of these new platforms.

The First Step

Before introducing a control plane, an honest inventory pays off: how many agents are currently running in your organisation — sanctioned and unsanctioned — and who actually knows all of them? This exercise can often be completed quickly but provides the foundation for every governance decision that follows.

Frequently asked questions

What is an AI Agent Control Plane?: An Agent Control Plane is a centralised control layer that combines visibility, access rights, runtime monitoring and cost management for AI agents across an organisation — comparable to a security and governance hub for every active agent.
Why did multiple vendors launch products on the same day?: CrowdStrike deliberately tied its launch to its own conference, Fal.Con. The launches from Boomi, Orchestry, Ping Identity, Kong and LogicGate were independent of that event — their coincidence shows that the entire industry is reacting, separately, to the same problem: agent sprawl in production.
What does AIDR (AI Detection and Response) mean?: AIDR is a new security category coined by CrowdStrike, specifically designed to detect and respond to risks created by AI agents — analogous to established categories like EDR for endpoints.
Does this matter for Swiss companies without US operations?: Yes. Once AI agents work with business or customer data, questions of access rights, traceability and accountability apply regardless of location. For companies with EU business, EU AI Act classification adds a further layer of relevance.
What is agent sprawl?: Agent sprawl describes a state in which more AI agents are active within an organisation than IT and business teams can track — often including self-built shadow agents with no central record.
Where should a company start if it has no governance in place yet?: With an inventory: which agents already exist, who created them, and which systems and data they touch. Only after that can you meaningfully decide which control-plane capabilities you actually need.

Sources

Would you like to explore this topic for your company?

Check Availability

AI and Economic Growth: What the 20% GDP Boost Means for Swiss Decision-Makers AI Agent or Better Processes? A Decision Framework for SMEs Okta Agent SSO: Enterprise Identity for AI Agents at No Extra Cost