Google named Leader in IDC agentic AI platform report

Google has been named a Leader in the IDC MarketScape: Worldwide Agentic AI Platforms 2026 Vendor Assessment, placing it among the leading suppliers in the newly defined market for agentic AI platforms.

The assessment measures vendors on short-term execution and longer-term strategy, using qualitative and quantitative criteria to map their market position. Google pointed to its Gemini platform as the basis for the ranking, arguing that customers are focused less on raw model performance and more on deployment, governance, integration and security.

That emphasis reflects a broader shift in enterprise AI buying. Large organisations are moving beyond pilots and asking whether AI agents can work across existing software, use company data safely and operate within established controls.

Platform focus

Google said Gemini is designed to work across the applications employees already use, including Google Workspace, Microsoft 365 and team chat tools. It said the system maintains the same memory, skills and permissions across desktop and mobile environments.

For developers, the platform spans no-code and low-code tools as well as code-first workflows through its Agent Development Kit and a tool called Antigravity. It also includes an Agent Registry intended to let teams publish and reuse workflows as shared skills across an organisation.

Google presented that approach as a way to connect business users and technical teams on a single platform, rather than splitting work across separate systems.

Model choice

Google also argued for openness across models, data sources and software tools. Through Model Garden, customers can access more than 200 models, including Google DeepMind models, Anthropic models and other third-party or open models, according to the company.

Gemini also connects to enterprise data in Google Cloud products such as BigQuery and Cloud Storage, as well as external systems including Salesforce and SAP, Google said. It added that the platform uses existing employee credentials so access controls remain tied to the original data source.

The platform is built on open protocols including MCP and A2A, which Google said support communication across different vendors' systems and existing toolchains.

Cost controls

Cost management has also come into sharper focus as companies test more advanced AI systems. Google said Gemini uses an automatic mode that selects models dynamically, defaulting to lighter models for routine work and using more advanced reasoning models only when needed.

Administrators can also apply spending limits by project, department or individual agent through the billing console, Google said. Under that setup, new requests pause when a budget threshold is reached, while active tasks can continue until completion.

Google linked this feature to what it described as smarter token economics, a term used to describe efforts to manage the cost, speed and model choice associated with AI workloads.

Security measures

Security and governance remain central concerns for companies considering broader use of AI agents, particularly where systems can act on behalf of staff. Google said each agent receives its own authenticated identity with least-privilege permissions, and that actions are recorded in an audit trail linked to both the agent and the user involved in the task.

It also said dynamic tasks run inside isolated sandboxes intended to limit exposure to corporate infrastructure and sensitive data. In addition, an Agent Gateway applies corporate security policies automatically and blocks unauthorised access or data leakage, according to Google.

These issues have become a key dividing line in the market as vendors try to persuade large organisations that AI agents can be used in production environments rather than only in limited tests.

"We are honored by IDC MarketScape's recognition of Google as a Leader, and we look forward to continuing this work alongside our customers," Google said.