How to Choose a Primary AI Model in Copilot Studio
The capabilities of artificial intelligence (AI) are rapidly evolving, and each generational model offers unique strengths, whether it's faster response times.
Browse more Learn programming through practical tutorials, code examples and clear guides covering languages, databases, web development and core concepts.
The capabilities of artificial intelligence (AI) are rapidly evolving, and each generational model offers unique strengths, whether it's faster response times.
An agent built from the Sustainability Insights agent template helps you uncover insights and data about your company's sustainability goals and progress.
Agents can use either generative or classical scheduling. By default, newly created agents will use generative scheduling.
An agent built from the Safe Travels agent template is a business-to-employee (B2E) agent designed to provide travel support for a company's employees.
An agent built from an IT Helpdesk agent template uses your organization's knowledge base to improve operational efficiency, enhance employee satisfaction, and.
An agent built from the Citizen Services agent platform allows public sector organizations to quickly build agents using their publicly available information to.
An agent built from the Financial Insights agent template allows users to build their own agents within Copilot Studio, and is designed for financial services.
Microsoft provides a catalog of managed agents that you can browse and install directly in Copilot Studio.
To help you get started building custom agents, Microsoft Copilot Studio provides several agent templates that you can use to create your own custom agents.
Create scoped Cursor Project Rules in .cursor/rules, configure all four activation modes, write valid frontmatter, troubleshoot matching, and share rules safely.
Generate an English dialogue with Educaplay's Ray assistant, refine the language and audio, assign a student role, test the activity, and share it with a class.
Explore running DeepSeek V4-Pro in Claude Code, including key concepts, practical steps, useful tips, and common questions.
Update a Copilot Studio test set's name, evaluation methods, test cases, user profile, and connections, or safely delete obsolete sets and results.
Edit single-response questions or multi-turn conversations, configure expected outcomes for each scoring method, and add cases to an existing test set.
Choose among general quality, meaning, tool use, keyword, similarity, exact-match, and custom checks based on the behavior each test case must prove.
Run a saved agent test set, investigate Pass, Fail, Invalid, and Error outcomes, compare two runs, and export results before they expire.
List Copilot Studio test sets, trigger an asynchronous evaluation, poll the run, retrieve case results, and integrate the checks into a release workflow.
Use Copilot Studio's test panel to trace conversation paths, inspect variables, manage connections, capture diagnostics, and confirm behavior before publishing.
Understand Copilot Studio agent evaluations, test cases, test sets, scoring methods, automation options, and where manual testing still matters.
Create a Copilot Studio single-response test set manually, from agent knowledge or themes, or by importing a CSV, then choose suitable evaluation methods.