MCP Use Cases for HR, Sales and Marketing
Sample MCP workflows for HR, Sales and Marketing, showing which parts are Tools, Resources and Prompts, and where humans must review.
Monday morning: HR opens the calendar to book interviews, sales reads customer history before a meeting, and marketing pulls campaign numbers from several dashboards into a deck. All three teams spend more time "going to get the data" than thinking. This is where MCP helps most clearly.
What is an MCP use case?
MCP is the standard that lets AI apps connect to systems through MCP Servers, which offer three capabilities: Tools for AI to act, Resources for context, and Prompts as templates the user invokes. The MCP documentation lists common Servers such as filesystems, databases, GitHub, Slack and calendars. A good use case arranges these pieces into a workflow with checkpoints for people.
HR
Scheduling interviews
- Resource: job details and the shortlist from a recruitment system that has an MCP Server
- Tool: find free slots in the interview panel's calendars and create the appointment
- Checkpoint: AI proposes times; HR confirms before invitations go out
Answering policy questions
When employees ask about benefits or leave, AI searches the policy handbook exposed as a Resource and answers with a reference to the document, cutting repeated questions to HR. If a question falls outside the handbook, AI should say it found no information and suggest contacting HR directly, rather than guessing.
Example
A factory in Chonburi hires dozens of production staff each month. HR uses a "/summarise-candidate" Prompt so AI summarises each applicant's experience against the job requirements, and HR decides who moves forward.
Sales
Preparing for customer meetings
- Tool: pull purchase history, open quotations and previous call notes from the CRM
- Prompt: "/meeting-prep", which summarises customer status, risks and questions to ask
Following up after meetings
AI drafts a summary email and updates the deal status in the CRM, but the rep must review before anything is saved or sent.
Example
A company selling POS systems to restaurants: before a visit, a rep asks, "How many branches use us, and what problems did they have last month?" AI pulls from the CRM and the support ticketing system and summarises it in bullet points within a minute.
Marketing
Campaign reports
- Tool: query the campaign results database by period and channel
- Resource: the team's metric definitions, so AI calculates the same way every time
Content planning
A Prompt preloaded with brand guidelines helps draft posts in a consistent voice no matter who on the team asks, and a tool adds them to the content calendar in the team's tool.
Example
A Thai beverage brand has AI pull last month's campaign results, compare them with targets, and draft a five-slide summary for the manager to review. The team gets more time to plan the next campaign, while deciding where to shift budget remains the manager's call.
Things to watch out for
- Applicant and customer data falls under PDPA. Consult your legal team or a qualified professional before giving AI access.
- AI should not make hiring decisions. Use it to summarise, but people must decide, to reduce the risk of bias.
- Check the numbers. AI may query the wrong period. Verify against the source before sending to executives.
- Separate permissions by role. Sales should never see salary data through an HR connector.
- External content can hide instructions, for example in résumés or customer emails. An AI reading such content should not have permission to send data out on its own.
MCP does not do the team's work. It stops the team wasting time fetching data.
How to get started
- Ask each team to list the three tasks that take the most data-gathering time each week
- Check whether the related systems offer an official MCP Server or need one built
- Map the workflow, separating Tools, Resources, Prompts and human checkpoints
- Pilot with one team for two to four weeks, then measure the time actually saved
Before connecting real data, read Using MCP safely with internal data, and HR teams should also read PDPA for HR.