Video Summary
Log into Copilot via m365copilot.com, Teams, or the Copilot app to access chat and agent creation.
Use Copilot Studio templates or start custom to prefill descriptions, instructions, and prompts.
For Copilot Chat users, ask the agent to request files as a workaround since non‑M365 licenses can only reference public URLs.
Enable code and image interpreters to generate charts and visuals, but toggle them on cautiously.
Test agents before creating, then share, edit, or delete via the agents dashboard.
No — Copilot Chat that comes with most Microsoft 365 business and enterprise accounts lets you build simple agents without additional licensing. More advanced features (like attaching SharePoint or private file sources) require an M365 Copilot license.
Copilot Chat users who can't attach private files can add an instruction in the agent to "ask me for the file." When the agent runs it will prompt you to upload the document, which acts as a practical workaround.
The code interpreter can analyze data and produce charts/graphs; the image interpreter can generate or interpret visuals. They add powerful capabilities but should be toggled on only when needed and tested for expected behavior.
Yes — Copilot Studio allows you to test an agent before creation. After creation you get a unique link to share, and you can edit, manage access, or delete agents from the "view all agents" dashboard.
Position Pal scans submitted résumés against a job posting, compares candidates, ranks them in a comparison table, and can produce visual summaries (like bar charts) that can be exported into documents.
"Today I'm going to show you how to build a simple and free co-pilot agent."
The video begins with an introduction to building a co-pilot agent, emphasizing that nearly all business and enterprise accounts already include co-pilot chat, which requires no additional licensing.
The presenter, Shervin Chaffy, encourages viewers to follow along and create their own agents, explaining that these agents can be powerful tools with the right tips.
"The first thing we're going to do is log in to Copilot."
Viewers are instructed to log into Copilot through various methods, including using a web browser at m365copilot.com or directly from Microsoft Teams.
Once logged in, users will see options for chat and creating an agent, prompting the discussion about the process to create a new agent.
"I'm just going to select create agent."
Shervin demonstrates the steps to create a new agent by selecting the "create agent" option, leading to the Copilot Studio interface.
A variety of templates are available for use, such as a career coach, customer insights idea coach, and a quiz tutor.
"These templates are examples of agents that you can use without any customization."
The presenter explains how to select and customize a quiz tutor template, highlighting the information that automatically populates in the interface.
Users can modify instructions and configurations while keeping the template's structure for efficiency.
"The main reason you'd want to make an agent is to handle repetitive tasks."
Shervin explains that agents are designed to simplify repetitive tasks that involve analysis and logic, effectively leveraging AI capabilities.
He further describes the limitations of Copilot chat regarding file attachments and provides a workaround by instructing users to ask for files within the agent's instructions.
"So, let's see if it can generate a quiz for us."
The process of generating a quiz is demonstrated as the agent prompts for a website URL to create questions.
An example quiz question is provided, showcasing the agent's ability to generate responses based on the given content accurately.
"I would suggest that you go and look at all the different templates."
Shervin encourages viewers to explore various available templates to create specialized agents that suit different needs, such as a writing or meeting coach.
The tutorial leads into creating a custom agent, emphasizing the flexibility of the Copilot Studio environment while highlighting the steps to start from scratch.
"I want to build an agent that scans résumés for a particular position and determines which candidates are the most qualified."
The video introduces the concept of an AI agent that helps automate the tedious task of reviewing résumés against job postings to identify qualified candidates. This agent is humorously named "Position Pal."
Users can customize the agent by uploading an icon or image; for example, the creator chose an image resembling Einstein for their agent.
The agent is characterized by a descriptive role where it reviews résumés and compares them to a job posting, effectively ranking the candidates based on various criteria.
"Compare résumés of candidates to the job posting and assess how they compare to the job."
The instructions for the Position Pal include a straightforward directive to evaluate résumés against the specified job posting, generating a comparison table with candidate ratings.
The agent can also prompt users to provide candidates' résumés and the job posting if they are not pre-attached, offering flexibility in how information is submitted.
Suggested templates can guide the tone of communication, such as maintaining a professional demeanor or incorporating humor.
"When you enable the code interpreter, it can create graphs and charts based on prompts."
The video highlights two powerful features of the agent: the code interpreter and the image interpreter, which can be toggled on or off depending on user needs.
The code interpreter can generate graphs and charts based on specific data inputs, while the image interpreter can create visuals from user prompts.
Although they can create additional functionalities, users are advised to enable these features cautiously and to edit the agent if issues arise.
"Would you like to test the agent before you create it?"
Before finalizing the agent's setup, viewers are encouraged to test its functionality using the prompts designed during setup, such as requesting résumés for evaluation.
After testing, users can create the agent, and a unique link is generated for sharing it within an organization or modifying its access permissions.
Once created, the Position Pal will appear under the user’s agents section, allowing for easy access and interaction.
"Please share with me the job and résumés for a comparison."
Users demonstrate how to interact with the Position Pal by providing a job posting and several résumés for evaluation. The agent responds by querying for required documents and acknowledges receipt of each.
Upon submission of the documents, the agent generates a detailed comparison, showcasing a structured table that evaluates each candidate's qualifications, highlighting differences in experience and skills.
The results include a summarized fit for each candidate, making it easier for employers or recruiters to identify the best applicants.
"This is something that you can copy, download, and insert into your document."
The agent can create visual representations of candidate qualifications, including bar charts rating various skills, which can be easily integrated into documents.
Users can add generated charts to pages, enhancing the presentation of data and providing consolidated reports that include both tables and visual components.
This functionality empowers users to export their findings into Word documents, facilitating the sharing of candidate evaluations in a professional format.
"You can always come back to it, edit it, or uninstall it if you need to."
Users can manage their AI agents by editing, uninstalling, or sharing them.
The editing feature allows users to modify instructions, adjust prompts, and make the agent perform better based on their needs.
By navigating to the view all agents section, users can delete any agents they no longer wish to keep.
"The main feature that they have that Copilot chat users don't is that you can actually add a lot of different knowledge sources."
M365 Copilot users have additional capabilities, such as attaching multiple knowledge sources like files, sites, and chats to their agents.
This enhances the agent's memory, allowing it to pull information from various sources for more robust performance.
For users without the M365 Copilot license, the workaround is still effective by prompting the agent for external sites.
"I want people to be empowered to create these agents."
The speaker encourages viewers to build their own custom AI agents, highlighting that everyone now has access to Copilot Chat without the need for an additional license.
The ability to create effective AI agents can lead to powerful productivity enhancements, even for those without M365 licensing.
The speaker references their playlist on advanced agents that require a license for more complex functionalities, inviting users to explore this content further.
