Artificial intelligence news.
Straight from the original announcements.

Google named a Leader in first IDC MarketScape for Agentic AI Platforms

October 19, 2026, 09:00Source: Google Cloud
Google named a Leader in first IDC MarketScape for Agentic AI Platforms
Image: Google Cloud

Google says it has been named a Leader in the inaugural IDC MarketScape: Worldwide Agentic AI Platforms 2026 Vendor Assessment. In a blog post, Google Cloud ties the result to Gemini and argues that deploying agents across a company, not raw model intelligence, is the main enterprise challenge.

Key points
  • Google says IDC named it a Leader in the inaugural IDC MarketScape: Worldwide Agentic AI Platforms 2026 Vendor Assessment.
  • Google frames the main enterprise challenge as deploying agents with quality, integration, security and governance, not raw model intelligence.
  • According to Google, Model Garden offers over 200 models, including Gemini 4 Argon and the latest Anthropic models from day one.
  • Auto mode picks a model per task step, and administrators can set budgets by project, department or agent.
  • Each agent gets its own identity and audit trail, and Google says governance comes at no additional cost.

What Google says about the IDC recognition

Google Cloud published the announcement on October 19, 2026, in a post by Jamie de Guerre, Senior Director, Outbound Product Management. According to Google, organizations keep telling it that the hard part of adopting AI is not model intelligence. The difficulty lies in deploying agents across the company while meeting requirements for quality, integration, security and governance.

Google says Gemini provides a connected foundation across the enterprise, and that this is why it believes it was named a Leader in the first IDC MarketScape on agentic AI platforms. The company describes the IDC model as an overview of the competitive fitness of technology and suppliers in a given market. It combines qualitative and quantitative criteria into a single graphic showing each supplier's position.

Two scores shape that position, according to the post. The Capabilities score measures a supplier's product, go-to-market and business execution in the short term. The Strategy score measures how well a supplier's strategy aligns with customer requirements over a 3-5 year timeframe. Market share appears as the size of each icon.

Four differentiators Google highlights

Google argues that a viable agentic platform has to meet work where it already happens and run under production-grade controls. It lists four core differentiators for Gemini.

  • Connected surfaces: Employees can work with Gemini inside Google Workspace, Microsoft 365 or team chats, with the same persistent memory, skills and enterprise permissions on desktop and mobile. Builders can choose from no-code and low-code tools to code-first workflows with the Agent Development Kit (ADK) and Antigravity. Teams can package a workflow as a shared skill and publish it to a central Agent Registry.
  • Openness: Through Model Garden, teams get access to over 200 models, including Google DeepMind models such as Gemini 4 Argon, day one access to the latest Anthropic models, and third-party and open models. Gemini connects to live data in BigQuery and Cloud Storage, as well as Salesforce, SAP and others, and inherits each employee's existing credentials. It is built on open protocols such as MCP and A2A.
  • Token economics: By default Gemini uses Auto mode, which matches each step of a task to a suitable model. It starts with a fast model like Gemini Flash and escalates to a high-capacity frontier reasoning model only when complexity requires it.
  • Governance and security: Google says governance is built in at no additional cost.

Cost controls and security details

On cost, Google says teams can reduce repetitive token consumption by packaging proven processes into shared skills. Administrators can set spending limits by project, department or individual agent in the billing console. When a threshold is reached, new requests pause while active tasks are allowed to complete, which Google says gives finance and engineering predictable cost controls without abruptly interrupting work in progress.

On security, every agent receives its own authenticated identity governed by least-privilege permissions. Each event is recorded in an audit trail tied to that identity and to the user who was using the agent. Isolated sandboxes keep dynamic tasks contained, and Agent Gateway applies corporate security policies, blocks data leakage and ensures agents only access what they are explicitly authorized to touch. Google contrasts this with relying on a model to self-police.

What it means for enterprise buyers

The post is a vendor's own reading of an analyst report, and the ranking details are in an IDC excerpt that Google invites readers to download. The source does not name other vendors in the assessment or give scores. For companies evaluating agent platforms, the announcement shows the criteria Google wants buyers to weigh: model choice, integration with existing data and tools, cost management and agent-level governance. Google says it looks forward to continuing this work with customers.

Related news

Coframe raises $35 million to build its self-improving ‘living internet’
BusinessNovember 9, 2026, 15:40

Coframe raises $35 million to build its self-improving ‘living internet’

Coframe has raised $35 million to develop an autonomous AI platform that continuously generates, tests and improves websites, landing pages and ads. The company says its…

Lovable acquires Radish AI team to expand business agent capabilities
BusinessNovember 9, 2026, 13:00

Lovable acquires Radish AI team to expand business agent capabilities

Lovable has acquired the team behind Radish AI, bringing its founders and senior engineering leaders into Lovable’s engineering organization to develop AI agents and…

Salesforce introduces Koa Decisions for faster enterprise AI choices
ModelsNovember 6, 2026, 17:00

Salesforce introduces Koa Decisions for faster enterprise AI choices

Salesforce has introduced Koa Decisions, a decision model that reads a request and a user-defined list of categories before selecting an answer in a single step. The…