Researcher, Analyst & Real-World Use Cases
Microsoft built specialist agents for knowledge discovery and data analysis. Knowing when to recommend Researcher, Analyst, a prebuilt agent, or a custom agent is a core exam skill.
Meet the specialist agents
Think of a hospital.
Copilot is the GP — good at everything, handles most problems, and refers you when needed.
Researcher is the radiologist — an expert at finding things hidden in mountains of data. Give it a pile of documents, emails, and web pages, and it’ll extract exactly what you need.
Analyst is the lab technician — give it numbers, spreadsheets, and datasets, and it’ll spot patterns, run calculations, and produce charts that tell a clear story.
Neither replaces the GP. They each handle the work the GP isn’t specialised for.
Researcher vs Analyst — the key comparison
| Feature | Researcher | Analyst |
|---|---|---|
| Works with | Unstructured data (docs, emails, chats, web) | Structured data (spreadsheets, databases, Power BI) |
| What it does | Finds, filters, summarises, and cites information | Analyses, calculates, visualises, and forecasts |
| Output | Structured reports with citations | Charts, visualisations, statistical insights |
| Think of it as | Research assistant / librarian | Data scientist / analyst |
| Best for | Market research, executive briefings, knowledge discovery | Budget analysis, trend identification, anomaly detection |
| Key technology | Semantic search across Microsoft Graph + web | Python reasoning, advanced calculations, Power BI integration |
Exam tip: The exam often presents a scenario and asks “which agent should they use?” The deciding factor is always: Is the data unstructured (documents, emails) → Researcher or structured (spreadsheets, numbers) → Analyst?
Researcher use cases
Researcher excels at turning scattered, unstructured information into clear, organised reports:
Scenario: Clearfield Council's compliance audit
Officer Patel needs to prepare a briefing on all data breaches reported in the last 12 months. The information is scattered across:
- SharePoint incident logs
- Teams channel discussions
- Email threads with external auditors
- Published guidance documents on the web
Without Researcher: Officer Patel spends 3 days manually searching each system, copying notes, and writing the report.
With Researcher: Officer Patel prompts: “Compile all data breach incidents from the past 12 months. Include root causes, departments affected, and resolution timelines. Cite all sources.”
Researcher searches across Graph data and the web, produces a structured report with citations, and Officer Patel spends 2 hours reviewing and refining — instead of 3 days.
Other Researcher use cases:
- Market research — “What are competitors saying about AI adoption in government?”
- Executive briefings — “Summarise everything discussed about Project Phoenix this quarter”
- Contract review — “Find all references to liability clauses across our uploaded contracts”
- Onboarding — “Compile a new-hire guide from our HR SharePoint site, training docs, and Teams FAQs”
Analyst use cases
Analyst excels at turning raw numbers into actionable business insights:
Scenario: Northwave's quarterly budget review
Alex (Northwave’s CEO) asks: “Where are we overspending compared to forecast?”
The data sits in 8 Excel workbooks across different departments, plus 2 Power BI dashboards.
Without Analyst: The finance team spends a week consolidating spreadsheets, running pivot tables, and building charts.
With Analyst: Alex prompts: “Compare actual Q3 spend vs forecast across all departments. Identify the top 3 areas of overspend, show the trend over the last 4 quarters, and suggest where to cut.”
Analyst processes the data, generates visualisations, and even provides a narrative: “Marketing overspent by 15%, driven by unplanned agency costs in August. Travel expenses dropped 22% vs forecast.”
Other Analyst use cases:
- Sales forecasting — “Based on the last 2 years, what should we expect in Q4?”
- Anomaly detection — “Flag any unusual transactions in this expense dataset”
- Power BI dashboards — “Break down shipping delays by region and supplier”
- Budget planning — “Model three scenarios: flat growth, 10% growth, 20% growth”
Other prebuilt agents
Beyond Researcher and Analyst, Microsoft provides additional prebuilt agents:
| Agent | What It Does |
|---|---|
| Prompt Coach | Helps craft better prompts — feedback, examples, and suggestions |
| Writing Coach | Gives constructive feedback on any writing task |
| Career Coach | Professional development advice, interview prep, skill planning |
| Ideas Coach | Brainstorming support — generates, ranks, and refines ideas |
These are lower stakes for the exam but worth knowing. They’re also available as templates when creating custom agents.
The decision framework — which tool for which problem?
| Tool | Best For |
|---|---|
| Copilot (in-app) | Interactive tasks in a single app — draft, summarise, analyse in context |
| Copilot Chat (Business Chat) | Cross-app knowledge search — 'What's the latest on X across all channels?' |
| Researcher | Deep unstructured research — compile reports from docs, emails, web |
| Analyst | Structured data analysis — trends, forecasts, visualisations from spreadsheets |
| Prebuilt agent | Common scenarios — writing help, career advice, brainstorming |
| Everyday user agent | Department Q&A bots, guided processes, simple lookups |
| SharePoint agent | Site-specific content discovery, document help, navigation |
| Custom advanced agent | Multi-step workflows, cross-system automation, API integrations |
Northwave's decision matrix — 5 problems, 5 tools
Each Northwave department has a different need. Here’s what Maya recommends:
- Sam (Marketing): “Help me write this blog post” → Copilot in Word (interactive, single-app)
- Jordan (CISO): “Compile all security incidents from the last year” → Researcher (unstructured data across multiple sources)
- Priya (Compliance): “Show me DLP violation trends by quarter” → Analyst (structured data, visualisations needed)
- New hires: “Where do I find the expense policy?” → SharePoint agent (site-specific content lookup)
- Finance team: “Auto-process 500 invoices monthly” → Custom agent (multi-step workflow, API integration)
🎬 Video walkthrough
🎬 Video coming soon
Researcher, Analyst & Use Cases — AB-900 Module 25
Researcher, Analyst & Use Cases — AB-900 Module 25
~10 minFlashcards
Knowledge Check
Clearfield Council needs to identify which departments had the highest variance between budget forecast and actual spending over the last 3 years. The data is in Excel spreadsheets and a Power BI dashboard. Which tool should they use?
Northwave's legal team needs to review 200 contracts and identify all clauses related to data processing agreements. The contracts are stored as Word documents in SharePoint and some are shared via email. Which tool is BEST suited?
Next up: Managing Copilot — the day-to-day admin tasks: assigning licenses, monitoring billing, tracking adoption, and governing prompts.