1. Executive Summary 

The Hiring Copilot Agent is an AI-powered recruitment automation solution built on Microsoft Copilot Studio, designed for staffing and recruitment agencies that manage large volumes of hiring requests on behalf of client companies. 

The agent monitors incoming Outlook emails for job hiring requests, extracts and interprets job descriptions using OpenAI (Claude Sonnet 4.6), and automatically matches requirements against a SharePoint-hosted candidate repository. It then shortlists the most suitable candidates, responds to the client’s email with a curated shortlist and attached CVs, and optionally schedules Microsoft Teams interviews with a single automated workflow. 

Built entirely on the Microsoft 365 ecosystem and orchestrated through MCP (Model Context Protocol) connectors, the agent eliminates manual email triage and CV screening, reducing recruiter effort from hours to seconds per hiring request. 

Key Outcome	Impact
Automated email monitoring and job description extraction	Eliminates manual inbox triage
AI-driven candidate matching against SharePoint repository	Reduces CV screening time by 85%+
Automated reply with shortlisted profiles and attached CVs	Faster, more consistent client responses
Automated Teams interview scheduling for clients and candidates	Removes back-and-forth scheduling overhead

2. Problem Statement 

2.1 Background 

Recruitment agencies act as intermediaries between companies seeking talent and a pool of registered candidates. Traditionally, a recruiter manually reads each inbound job request email, interprets the requirements, searches through spreadsheets or resume folders, shortlists candidates, and sends replies, a process repeated dozens of times per day. 

2.2 Core Problem 

There is no automated system to connect inbound hiring requests with the most relevant candidate profiles. Every step, reading emails, interpreting JDs, searching CVs, drafting replies, and scheduling interviews, is done manually, leading to bottlenecks, inconsistency, and slow turnaround times. 

2.3 Affected Stakeholders 

Stakeholder	Role	Impact	Key Concerns
Recruitment Consultants	Process hiring requests daily	High	Speed, accuracy, workload reduction
Client Companies	Submit job requirements	High	Quality of shortlist, response time
Candidates	Receive interview invitations	Medium	Timely communication, fair evaluation

2.4 Pain Points 

• Recruiters spend 4–6 hours/day manually reading and routing hiring request emails 

• CV matching is subjective and inconsistent across recruiters 

• Candidate shortlists are often delayed, resulting in poor client experience 

• Interview scheduling involves multiple back-and-forth email exchanges 

3. Solution Overview 

3.1 Proposed Solution 

The Hiring Copilot Agent automates the end-to-end recruitment response pipeline. When a hiring email arrives in a designated Outlook inbox, the agent triggers via Power Automate, reads the email content through the Outlook MCP connector, and passes the job description to OpenAI for structured extraction (role title, required skills, experience level, location, etc.). 

The agent then queries the SharePoint candidate repository via the available SharePoint knowledge source, retrieving candidate profiles. Claude Sonnet 4.6 evaluates each candidate against the extracted job requirements using the following criteria:  
skill and experience alignment, depth and relevance of professional history, real contributions vs. generic responsibilities, technical fit and transferable skills, and overall career trajectory. 

The top-matched candidates are shortlisted, their CVs are fetched from SharePoint, and a structured reply email is composed and sent via the Outlook MCP connector. Optionally, the agent generates Microsoft Teams meeting links and sends calendar invitations to both the client and the shortlisted candidates. 

3.2 Goals & Success Metrics 

Goal	Success Metric	How Measured	Timeline
Email-to-shortlist speed	Under 2 minutes per request	Agent execution log timestamps	Month 1
Matching accuracy	90%+ recruiter approval rate on shortlists	Recruiter feedback loop	Month 1
Client response rate	100% of hiring emails receive automated reply	Outlook delivery reports	Month 1
Interview scheduling	Teams invites sent within 5 mins of shortlist approval	Calendar audit log	Month 1
Recruiter time savings	4+ hours/day reclaimed per consultant	Time-tracking comparison	Month 1

4. Workflow Overview 

4.1 End-to-End Workflow 

Step	Action	Description	System
1	Email Trigger	New hiring request email arrives in Outlook inbox; Power Automate detects it	Outlook + Power Automate
2	Email Reading	Agent reads email subject and body via Outlook MCP connector	Outlook MCP
3	JD Extraction	OpenAI extracts structured job requirements (role, skills, experience, location)	Claude Sonnet 4.6
4	Candidate Retrieval	Agent fetches candidate profiles from SharePoint repository via SharePoint Knowledge	SharePoint Knowledge Source
5	AI Matching	Claude Sonnet 4.6 scores and ranks candidates against extracted JD criteria	Claude Sonnet 4.6
6	CV Fetching	Top-matched candidates’ CVs are retrieved from SharePoint document library	SharePoint MCP
7	Reply Composition	Agent drafts and sends a structured email with shortlisted profiles and attached CVs	Outlook MCP
8	Interview Scheduling	Teams meeting links generated; calendar invites sent to client and candidates	Calendar MCP

5. High-Level Architecture 

5.1 Architectural Layers 

Layer	Technology	Responsibility
Orchestration Layer	Microsoft Copilot Studio + Power Automate	Agent logic, workflow triggers, MCP tool invocation
AI / Intelligence Layer	Claude Sonnet 4.6	JD extraction, candidate evaluation, reply drafting
Email Integration	Outlook / Exchange via MCP Connector	Read incoming emails, send shortlist replies
Candidate Data Layer	SharePoint Online via MCP Connector	retrieve candidate profiles and CVs
Scheduling Layer	Microsoft Teams via MCP Connector	Generate meeting links, send calendar invitations
Automation Layer	Power Automate	Event-driven triggers on new email arrival

5.2 Data Flow 

#	From	To	Description
1	Outlook Inbox	Power Automate	New email event trigger fired on arrival of hiring request
2	Power Automate	Copilot Studio Agent	Email metadata and body passed to agent for processing
3	Agent	Outlook MCP	Full email content retrieved for analysis
4	Agent	Claude	Raw email body sent for job description extraction
5	Claude	Agent	Structured JD object returned (skills, role, seniority, location)
6	Agent	SharePoint MCP / Knowledge	Query candidate profiles matching extracted criteria
7	SharePoint MCP	Agent	Candidate profile list returned with metadata
8	Agent	Claude	Profiles + JD sent for comparative scoring and ranking
9	Claude	Agent	Ranked shortlist with evaluation rationale returned
10	Agent	SharePoint MCP	CV documents for top candidates fetched
11	Agent	Outlook MCP	Reply email with shortlist and CVs composed and dispatched
12	Agent	Calendar MCP	Meeting links generated; invitations sent to client and candidates

Candidates Requirement Email: 

SharePoint Location from where the Agent will pick the Resumes: 

MCP’s Used in the Agent: 

Response from the agent including detailed candidate comparison word document as well as the Resumes of the Candidates: 

Requesting to arrange the interviews with the candidates: 

Agent schedules interviews with the candidates by extracting their emails from the resumes: 

5.3 Model Choice & Rationale 

Attribute	Value	Notes
Selected Model	Claude Sonnet 4.6	Primary intelligence engine for extraction and matching
Model Type	Large Language Model Generation + Reasoning	Handles both structured extraction and nuanced evaluation
Key Capability	Function calling / structured outputs	Ensures reliable JSON extraction from unstructured emails

5.4 MCP Integrations 

MCP Connector	Auth Method	Data Exchanged
Outlook MCP	OAuth 2.0 / Azure AD	Read emails, send replies with attachments
SharePoint MCP	OAuth 2.0 / Azure AD	Query lists, fetch documents and CVs
Microsoft Calendar MCP	OAuth 2.0 / Azure AD	Create online meetings, send calendar invites

6. Deployment & Infrastructure 

6.1 Environments 

Environment	Platform	Purpose
Development	Copilot Studio Dev Environment	Feature building and unit testing
Staging	Copilot Studio Test Environment	Integration testing with real M365 connectors
Production	Copilot Studio Production Environment	Live recruitment automation

7. Open Issues & Risks 

Risk	Severity	Likelihood
Low candidate match quality if profiles not regularly updated	High	Medium
Email parsing failures on non-standard JD formats	Medium	Medium
Claude Sonnet 4.6 producing biased candidate rankings	High	Low

8. Appendix 

A. Glossary 

Term	Definition
MCP	Model Context Protocol a standard for connecting AI agents to external tools and data sources
Copilot Studio	Microsoft’s low-code platform for building AI-powered copilot agents
Claude Sonnet 4.6	OpenAI’s multimodal frontier model
Power Automate	Microsoft’s workflow automation platform used to trigger and orchestrate agent actions
JD	Job Description the structured requirements document extracted from hiring emails

Read more : Power Automate Solutions for Growing Businesses

FAQ’s