ETA Hot Topic Hub · AI Agents

AI agents are clocking in. Is your team ready?

AI agents now hold roles, keep working after you log off and act inside everyday apps. Adoption is climbing fast, but value, security and management practices are lagging. ETA tracks enterprise adoption, agent platforms, security and governance, and what agents mean for the workforce.

Read the feature See the latest news

Updated November 8, 2026 · Stats checked weekly, stories added daily

40%Respondents at organizations above $1B revenue that report scaling AI agents in 2026, up from 27% a year earlier (McKinsey)Source

15xYear-over-year growth in active agents in the Microsoft 365 ecosystem, March 2025 to March 2026 (18x in large enterprises)Source

13%CIOs who say their enterprise has realized significant or transformational value from AI so far (Gartner, 2,002 CIOs, 2026)Source

2 in 3IT decision-makers who say most AI agents in their organization lack formal IT approval (Workato survey of 1,001, Oct 2026)Source

ETA analysis

Read the analysis

Illustration of an AI agent hub connected to work apps

Feature · November 8, 2026AI Agents Are Joining the Workforce. Most Organizations Are Not Ready to Manage Them.Agents now hold roles, keep working after you log off and act inside everyday apps. Adoption is surging, but value, security and management practices are lagging behind.Read the feature

Newsfeed

Latest news

Curated by ETA, each story with our take on what it means for leaders adopting AI.

Follow by RSS JSON feed Get it by email

AdoptionAnthropicOct 27, 2026
Anthropic expands GSA deal, extending discounted Claude to state and local governments (opens on Anthropic)
ETA take: Procurement barriers for state and local agencies are falling. Governance capacity and staff readiness are now the real constraint.
AdoptionGartnerOct 19, 2026
Gartner: only 13% of CIOs report significant value from AI so far (opens on Gartner)
ETA take: Leaders getting results set rules for scaling pilots and for shutting down low-value AI. Volume of agents is not a success metric.
SecurityCIO DiveOct 13, 2026
IT leaders brace for agent-driven security incidents as most agents run without formal approval (opens on CIO Dive)
ETA take: Shadow agents are the new shadow IT. Start with an inventory: you cannot govern agents you do not know exist.
PlatformsGoogle CloudOct 8, 2026
Google unveils Gemini agent with per-agent identity, audit trails and spend caps (opens on Google Cloud)
ETA take: Identity, least privilege, audit and cost caps are becoming standard agent features. Make them requirements in every agent purchase.
WorkforceAnthropicOct 2, 2026
Anthropic commits $100 million to train 10,000 engineers to deploy AI in enterprises (opens on Anthropic)
ETA take: Deploying and supervising agents is becoming a profession. Organizations should be building that skill set in house now.
PlatformsSalesforceSep 28, 2026
Salesforce launches Agentforce Coworker and long-horizon agents at Dreamforce 2026 (opens on Salesforce)
ETA take: Agents that pursue goals over weeks need approval checkpoints and monitoring. Ask vendors how oversight works before you ask what agents can do.
PlatformsOfficial Microsoft BlogSep 25, 2026
Microsoft introduces Autopilot, a Copilot teammate with its own identity and workspace (opens on Official Microsoft Blog)
ETA take: Agents now get identities, memory and usage-based billing. Budget owners and security teams both need a seat at the table.
SecurityThe Boston Globe (The New York Times)Sep 24, 2026
Australia says an OpenAI agent accessed non-public parts of a government health statistics portal (opens on The Boston Globe (The New York Times))
ETA take: The first widely reported case of an agent working around blocks on a government site. Agent risk is now a board-level issue, not a lab curiosity.
AdoptionMcKinsey & CompanyAug 25, 2026
McKinsey State of AI 2026: large companies scaling agents jumps to 40%, high performers stay rare (opens on McKinsey & Company)
ETA take: Big firms are scaling agents, but the share capturing real EBIT impact has not moved. Deployment alone does not create value.
AdoptionCIO DiveAug 12, 2026
Deloitte finds only 15% of organizations have scaled multiagent systems (opens on CIO Dive)
ETA take: Layering agents on old processes is fast but shallow. The payoff comes from redesigning the work itself, which takes years, not quarters.

Go further with ETA

Put this to work

AI Agents at Work: FoundationsFree ETA course: what agents are and how to use them safely at work.AI Agents at Work: PractitionerBuild and supervise agents for real team workflows.ETA credentialsEarn the Certified AI Empowered Professional credential.National AI AcceleratorBring a real agent project and get it to results.

FAQ

Common questions

What is an AI agent, and how is it different from a chatbot?

A chatbot answers questions; an agent takes actions. NIST describes agents as systems capable of autonomous actions, and 2026 products like Microsoft's Autopilot and Google's Gemini agent can hold a role, run multi-step work in other apps and keep going for hours or days without a prompt.

How many organizations are actually using AI agents at scale?

Fewer than the hype suggests. McKinsey's 2026 survey found 40% of respondents at organizations above $1 billion in revenue report scaling agents, but Deloitte found only 15% of organizations have scaled multiagent systems, and Gartner found just 13% of CIOs report significant or transformational value from AI so far.

What are the biggest security risks with AI agents?

Agents act with real permissions, so the risks include unapproved agents, over-broad access and agents taking actions nobody intended. In a Workato survey of 1,001 IT decision-makers, nearly two-thirds said most agents lack formal IT approval, and Australia reported in September 2026 that an OpenAI agent accessed non-public parts of a government health statistics portal.

Is government setting standards for AI agents?

Yes. NIST launched an AI Agent Standards Initiative in February 2026 focused on industry-led standards, interoperable protocols and research on agent identity, authorization and security. Its National Cybersecurity Center of Excellence also released a draft concept paper on identity and authorization for software and AI agents.