Claude models and the Claude Agent SDK for custom agents.
Agents.
Understand what agents can do.
Evaluate platforms, costs, and controls.
Platforms & providers
Explore tools for building custom agents and using agents in business software. Evaluate the product against the work you need it to do.
Agent development, deployment, and management on Google Cloud.

Copilot Studio and Foundry Agent Service support different ways to build and operate agents.
APIs and the Agents SDK for building agents with tools and handoffs.
Agentforce for sales, service, and other business workflows.
AI Agent Studio for building and testing agents in ServiceNow.
Featured providers, listed alphabetically. The categories overlap; this is a starting point for research, not a ranking. Profiles may require an account.
View the Institute assessment ↗Research for the buying team
Define the task, set your requirements, and decide what evidence a platform must provide before you commit.
Enterprise Agent Platforms
Vendor profiles, evaluation criteria, and a workbook for building your shortlist.
View the buyer’s guide ↗Account requiredThe platform decision starts with the work
Set the task, required connections, and approval limits before comparing products.
Measure AI economics at workflow level
Count the full cost of a usable result, including failed attempts, retries, and human review.
Start with a shared understanding.
Plain-language guides to what agents are and how to monitor their work. Start here before evaluating products or planning a pilot.
What are AI agents?
Understand how agents choose actions, where they differ from fixed automation, and where human control belongs.
Read the guideMonitoring & evaluationWhat is agent observability?
Learn how to trace an agent’s actions, investigate failures, and measure the cost of completed work.
Read the guide