Early Access
ActivTrak MCP is in Early Access (EA), with additional capabilities planned before General Availability. It's included at no additional cost with Workforce Management, Productivity Optimization, and the Workforce Management + Productivity Optimization Bundle.
ActivTrak MCP connects your AI assistant to your ActivTrak workforce data so you can ask questions in plain language and get answers, charts, and trend analyses without leaving the application. There's no CSV export to wrangle, no report to build manually, and no BI or developer request to submit.
Contents
- What is ActivTrak MCP?
- Why use ActivTrak MCP?
- What can you do with it?
- Use cases supported at launch
- Who can use it?
- Choose your AI assistant
- Get started
- Best practices
- Data access, privacy and guardrails
- Frequently asked questions
- Learn more
What is ActivTrak MCP?
MCP stands for Model Context Protocol, an open standard that lets AI applications securely connect to other business applications. ActivTrak MCP is a hosted server that exposes your ActivTrak workforce data to a connected AI application as tools it can use to answer your questions.
The work is split between two systems:
- ActivTrak supplies the data and enforces its existing calculation logic, data definitions, compliance guardrails, and user access settings.
- Your AI assistant provides the conversation, writes the response, and builds any visuals.
Because ActivTrak does the calculations, the numbers you get back follow the same definitions you see in the ActivTrak app.
Why use ActivTrak MCP?
ActivTrak gives you several ways to work with your data: reports and dashboards in the web app, BI tools such as Power BI or Tableau through ActivConnect, and now your own AI application. ActivTrak MCP is built for the moments when:
- An ad hoc question doesn't fit an existing dashboard. You'd otherwise export data and piece reports together by hand.
- The answer spans multiple systems. Combining ActivTrak data with other sources (CRM, HRIS, etc.) has typically meant a dedicated analyst or BI team.
- You want your own AI to work with ActivTrak data. Until now, there was no way to connect it.
With ActivTrak MCP, any authorized user can ask a question and get an answer backed by ActivTrak's logic, without analyst skills.
What can you do with it?
- Ask freeform questions. Type a question the way you'd ask a colleague, such as "Does the marketing team have available capacity to take on an additional messaging project?"
- Generate visuals and custom analyses. Ask for charts, trend lines, dashboards and multi-dimensional breakdowns, as your assistant supports them. These go beyond what dashboards and Work Advisor produce today.
- Combine data sources. AI assistants can connect to multiple MCP servers at once, so you can pair ActivTrak data with other systems your organization uses, such as Jira, Salesforce, or an HRIS. Ask about utilization alongside project status or deal activity in one conversation.
- Make decisions with confidence. The data you get through your AI assistant matches the data in the ActivTrak app and uses the same calculation logic. Your established permissions apply too, so Viewable Groups are honored, and users see only what they're supposed to see.
Use cases supported at launch
Here are some of the ways teams use ActivTrak MCP to turn workforce data into answers. We add new use cases every quarter, so check back as ActivTrak MCP grows.
| Use case | What it helps you do | Example questions |
|---|---|---|
| Staffing and capacity planning | Evaluate whether teams have room for new work and where capacity could be shared |
"Does the marketing team have available capacity to take on an
additional project?" "Which teams have capacity we could reallocate?" |
| Workforce utilization | Spot over- and under-capacity teams, burnout risk and redistribution opportunities |
"Which teams are under capacity this month?" "Who can absorb workload without overtime risk?" |
| PTO and attendance validation | Find unreported absences, PTO mismatches and compliance gaps |
"Who has taken days off without approved PTO?" "Which teams have the most unsubmitted PTO?" |
| Schedule adherence | Flag late starts, early finishes, long breaks and low adherence trends |
"Which teams have schedule adherence below 80%?" "Which team members are chronically late or leave early?" |
| Technology usage | Understand application and category usage and technology adoption patterns |
"Which tools is the Product team using most?" "Where does Engineering split time across tools?" |
| Potential false activity | Surface activity patterns that may indicate simulated input and review the activity around a flagged event |
"Show me all potential false activity concerns across my organization." "Which users have potential false activity flagged this month?" |
Who can use it?
ActivTrak MCP is included at no additional cost with Workforce Management, Productivity Optimization and the Workforce Management + Productivity Optimization Bundle. It’s not included with Work Activity Tracking accounts. No minimum ActivTrak Agent version is required.
| Who | What they do |
|---|---|
| ActivTrak Admin | Enables MCP access for your organization in ActivTrak |
| AI assistant administrator | Adds the ActivTrak MCP connector for your organization if your assistant requires it |
| Each user | Connects with their own existing ActivTrak credentials |
Choose your AI assistant
Setup steps differ by assistant. Select the guide for the one your organization uses.
| AI assistant | Setup guide |
|---|---|
| Claude (Anthropic) | Setup Guide: ActivTrak MCP for Claude |
| ChatGPT (OpenAI) | Setup Guide: ActivTrak MCP for ChatGPT |
| Copilot (Microsoft) | Coming soon |
| Gemini (Google) | Coming soon |
Get started
- Enable MCP access in ActivTrak. An ActivTrak Admin goes to APIs & Integrations > MCP Access, clicks Enable MCP access and confirms. No one in your account can connect until you do this.
- Set up your AI assistant. Follow the setup guide for your assistant. If your assistant requires an administrator to add the connector, that person completes it once for the organization.
- Connect your account. Each user signs in with their ActivTrak credentials, including multi-factor authentication if it's enabled on their account.
- Start asking. Return to your assistant and try a question from the use cases above.
Best practices
Write clear, specific questions
- Name the time frame. "Over the last 30 days" gives a better answer than "recently."
- Name the team or group. Specify who you want to look at, within the groups you're allowed to view.
- Name the metric. Ask for productive time, utilization, or schedule adherence rather than "performance."
- Say how you want the answer. Ask for a chart, a table or a short summary.
Start broad, then narrow
Begin with a team-level question, then ask follow-ups to explore a change. For example, ask which teams are over capacity, then ask which individuals on the team are driving it.
Compare against a baseline
A single number rarely tells you much. Ask for comparisons, such as this month compared to last month or one team compared to the organization overall.
Verify important numbers
Before you share a figure in a report or decision, spot-check it against the matching dashboard in ActivTrak. If a question can't be answered reliably, ActivTrak MCP explains why and suggests a rewording rather than returning a fabricated or blank response.
Combine ActivTrak data with other context
If you've connected other systems to your assistant, ask questions that bring them together. Workforce data is most useful when paired with the business context behind it.
Use insights to inform conversations
Responses are informational and advisory. Use them to spot patterns, ask better questions, and prepare for coaching or planning conversations with the people involved. They aren't a substitute for your own judgment.
Admins: review access regularly
- Use the connected users table on the MCP Access page to see who has connected, which AI assistant they use and when they last queried.
- Revoke a user's connection at any time from the same page.
- Remind users that they can see only what their ActivTrak role and Viewable Groups allow.
Data access, privacy and guardrails
- Your existing permissions apply. Secure authentication tied to your ActivTrak account controls access. The system automatically enforces each user's role permissions and Viewable Groups at sign-in. The AI assistant does not grant additional access.
- Out-of-scope requests are declined. If you ask about people outside your Viewable Groups, the request is declined with a clear explanation and no unauthorized data is shown.
- ActivTrak's guardrails still apply. The same compliance guardrails as the ActivTrak app apply here, and ActivTrak's calculation logic can't be bypassed. Responses are advisory. ActivTrak MCP doesn't provide termination, disciplinary, or individual performance recommendations.
- Queries are logged without employee data. Each query is logged with the account, timestamp, data domain, and success or fallback status. Logs contain only query metadata; we don't retain employee PII, and we encrypt logs in transit and at rest.
Frequently asked questions
Does ActivTrak MCP cost extra?
No. It's included at no additional cost for accounts on eligible packages: Workforce Management, Productivity Optimization, and the Workforce Management + Productivity Optimization Bundle.
How is ActivTrak MCP different from Work Advisor?
Work Advisor is ActivTrak's own AI assistant and answers questions inside ActivTrak. ActivTrak MCP gives your own AI assistant access to ActivTrak data, so you can build visuals and trend analyses and combine ActivTrak with other connected sources in a single answer.
Which AI assistants are supported?
See Choose your AI assistant for current availability.
Can I see data for people I don't have permission to view?
No. ActivTrak enforces your role permissions and Viewable Groups.
What happens if my assistant can't answer a question?
ActivTrak MCP explains in plain language why it can't answer and suggests a way to reword the question. It doesn't return a fabricated or blank response.
Can my admin see what I ask?
Admins can see which users have connected, which AI assistant they use, and when they last queried data. Query logs contain metadata only and no employee PII.
How do I disconnect my assistant from ActivTrak?
Disconnect from the MCP Access page in ActivTrak. You can reconnect at any time by following your assistant's setup guide. Admins can also revoke a user's connection.
How do I report an inaccurate or unexpected response?
Contact your ActivTrak Support or Customer Success team with details. Feature feedback can also go to your Customer Success team.