Google Cloud consolidated Vertex AI into the Gemini Enterprise Agent Platform this year and shipped the governance layer enterprises have been missing: Agent Identity gives every agent a unique cryptographic ID, Agent Gateway enforces runtime policy on every agent-to-agent and agent-to-tool call, and Model Armor guards against prompt injection and data leakage. The infrastructure is real and it is live. The adoption gap sitting underneath it is bigger than most executives realize.

A 2026 OutSystems survey of 1,900 IT leaders found 96% of enterprises already run AI agents in production. Only 12% say they can govern them. Separate identity-security research puts the scale of the problem in sharper terms: non-human identities, including AI agents, now outnumber human identities roughly 45 to 1 inside the average enterprise, 92% of organizations lack full visibility into those identities, and mean monitoring coverage across production agents sits at 52%, meaning close to half of all agents in production today run unmonitored. Seventy-eight percent of organizations have no documented policy for creating or retiring an AI identity. Eighty-four percent could not currently pass a compliance audit focused on agent behavior or access.

OpenAI reports enterprise ChatGPT and Codex growth tied to its GPT-5.6 series, and launched OpenAI DeployCo, a majority-owned consulting subsidiary with more than $4 billion in initial investment, alongside its acquisition of applied-AI consultancy Tomoro, a direct move into systems-integration territory. Anthropic shipped Claude Opus 5 and is independently tracked as the current leader in enterprise usage share; it did not join NVIDIA's Open Weights letter, which has grown to roughly 50 signatories without Amazon or Anthropic. NVIDIA is reportedly weighing a financing guarantee near $250 billion tied to an OpenAI data center project in Ohio and has invested in Safe Superintelligence.

On the connective-tissue layer, Google, Microsoft, Salesforce, Snowflake, ServiceNow, Cisco, Databricks, GitHub, NVIDIA and Hugging Face backed a new Apache 2.0 standard called Agent Resource Discovery, positioned as an alternative path to Anthropic's Model Context Protocol, which sits at roughly 97 million monthly SDK downloads. Separately, a 28-plus member Open Secure AI Alliance, spanning Adobe, Capital One, Databricks, IBM, Microsoft, NVIDIA, Salesforce, SAP, ServiceNow and Snowflake, formed around open-weight model security after a Hugging Face breach where open-weight tooling reportedly aided forensic remediation faster than closed-source alternatives. Read this as ecosystem hedging on the standards layer, not a verdict on any single protocol.

Snowflake shipped Cortex AI Gateway specifically to control agent sprawl and prevent runaway AI spend, the same cost-then-risk sequencing seen at AWS on Bedrock in the prior edition of this scan, now appearing at the data-platform layer too.

Financial Services

Forty-four percent of finance teams expect to run agentic AI in 2026, up from roughly 6% in 2025. Fifty of the world's largest banks announced more than 160 agentic use cases in 2025, led by real-time compliance monitoring replacing periodic review. KPMG documents an average 2.3x return on agentic AI investment within 13 months, with top performers earning $8 for every $1 invested. EY's 2026 read: institutions pairing agentic AI with strong governance are demonstrating regulatory leadership, not attracting extra scrutiny.

Healthcare

WellSpan Health's 13-month deployment of an AI voice agent logged close to 2 million patient conversations across outreach and inbound call management, evidence agentic AI is operating as a system-wide capability rather than a pilot. More than 80% of healthcare executives in a 2026 Deloitte survey expect agentic AI to deliver significant clinical and back-office value. AI scribe tools are cutting physician charting time 40 to 45 percent and clinical note error rates 25 to 30 percent among adopters.

Manufacturing

Named production deployments now include Siemens, Bosch, GE Vernova and Schneider Electric, spanning engineering productivity, vision-based quality inspection, grid decision support and multi-site energy optimization. Companies with agents in production report roughly 34% average gains in both production efficiency and supply-chain efficiency.

Energy & Utilities

Energy monitoring and optimization ranks among the top-three near-term AI priorities for about 40% of manufacturers surveyed, and utility-scale deployments from GE Vernova and Schneider Electric show capital moving ahead of governance maturity, the same identity and access gap seen in finance and healthcare, with higher physical-safety stakes attached.

EU AI Act Article 50 transparency obligations bind August 2, 2026, three days from this edition. Unlike most of the Act, Article 50 applies instantly to all in-scope systems regardless of when they were deployed, covering AI-interaction disclosure, AI-generated content labeling, emotion-recognition and biometric-categorization notice, and deepfake or synthetic public-interest content labeling. Penalties run up to €15 million or 3% of global annual turnover.

California's AI Transparency Act also becomes operative August 2, 2026, the identical date as the EU's Article 50. Any enterprise serving both EU and California users faces two independent transparency regimes activating on the same day.

Texas's Responsible AI Governance Act (HB 149) has been in force since January 1, 2026, currently the broadest in-force US state model. Colorado's rebuilt SB 26-189, signed May 14, 2026, now starts January 1, 2027 for covered automated decision technology.

The World Economic Forum finds more than 1 in 3 young workers globally sit in occupations with medium-to-high AI task exposure, and employers expect 39% of workers' core skills to change by 2030. Deloitte's 2026 Global Human Capital Trends report finds only 6% of leaders say they are making real progress designing how humans and AI actually work together day to day. McKinsey and PwC data show the most AI-exposed companies have tripled their productivity-growth lead since 2022, with the top quintile of AI-exposed firms averaging 163% productivity growth. US job postings requiring AI skills grew 144% year over year as of April 2026, and workers with demonstrable AI competencies earn roughly 56% more than peers without them.