ActivTrak MCP Server Tutorial: Connect Workforce Data to Claude and ChatGPT

Introduction

ActivTrak knows where the workday actually goes, which applications teams live in, and where capacity is stretched thin. Until recently, turning that knowledge into an answer meant opening dashboards, exporting data, or asking someone to build a custom report. ActivTrak MCP changes the route: it connects that behavioral work data to Claude and ChatGPT through the Model Context Protocol, so a manager can ask a plain-English question and get an answer grounded in real work patterns. This tutorial walks through the full setup, from confirming your plan is eligible to asking your first capacity question and verifying the answer against the dashboard. It is written for team leads, operations managers, and ActivTrak administrators who want AI answers they can trust, not just AI answers that sound confident.

Quick answer

Ask your ActivTrak administrator to enable ActivTrak MCP for Claude or ChatGPT, connect it as a remote MCP connector, then ask productivity questions in plain English and check each answer against the same date range in your ActivTrak dashboard.

Evidence

ActivTrak MCP is very new, so the evidence base is the vendor’s own launch material rather than independent benchmarks, and this tutorial says so plainly. ActivTrak introduced ActivTrak MCP on October 13, 2026, in Early Access for Claude and ChatGPT, with support for Microsoft Copilot and Google Gemini planned for later releases. More than 9,500 organizations worldwide use ActivTrak, according to the company’s announcement, which means the MCP launch lands on an already large installed base rather than a niche product. ActivTrak’s own 2026 workplace research found that 80 percent of employees used AI tools at work, a figure that explains why the company is connecting work data to AI instead of asking leaders to adopt yet another dashboard.

The vendor’s framing is worth quoting directly, because it states the exact problem this tutorial solves. “Many organizations have invested millions in AI and expect a meaningful return. But realizing that value requires AI to work across the business, drawing on data that lives in different systems,” said Javier Aldrete, Chief Product Officer at ActivTrak. “Leaders shouldn’t have to export data, request a custom report or switch applications every time a question spans multiple data sources. ActivTrak MCP provides the work intelligence to support that analysis, helping customers get more value from the AI and data they already have.”

Two design facts from the announcement shape everything below. First, account administrators control whether ActivTrak MCP is enabled and which AI platforms are allowed, and they can review usage and export an audit trail for compliance review. Second, ActivTrak MCP carries existing user permissions and privacy controls into AI analysis, so the AI cannot see data the signed-in user could not already see. If you remember only one governance rule from this tutorial, make it that one: MCP access follows the same visibility rules as the dashboard, it does not bypass them. For background on the protocol itself, the Model Context Protocol introduction explains how clients and servers exchange tools and data, and ActivTrak’s own company FAQ confirms that the ActivTrak MCP Server connects approved AI tools and agents to ActivTrak data and intelligence.

Prerequisites

• An ActivTrak account on the Workforce Management or Productivity Optimization plan, with the ActivTrak agent already deployed to employee devices, because no workforce analytics data appears until agents are installed.
• Administrator access to your ActivTrak account, or a cooperative administrator, because enabling MCP and choosing allowed platforms is an admin-controlled setting.
• A Claude or ChatGPT account that supports remote MCP connectors. Copilot and Gemini are not supported yet, so do not plan a rollout around them.
• A concrete first question in mind, such as a capacity concern for one team over the last four weeks, so you can verify the AI answer against a dashboard you already understand.
• Basic familiarity with MCP concepts helps but is not required. If the term is new, start with our MCP server tutorial, which explains what a server, a client, and a tool call are.

Step-by-step guide

Step 1: Confirm your plan is eligible

Before touching any settings, confirm that your account qualifies. ActivTrak MCP Early Access is available to customers on the Workforce Management and Productivity Optimization plans, and it is included at no additional cost for eligible plans during Early Access. If your account is on a different plan, the MCP option will not appear, and no amount of reconnecting will make it appear. Ask your administrator to confirm the plan, or check with ActivTrak support about Early Access availability for your account. This check takes two minutes and prevents the most common setup failure, which is troubleshooting a feature that was never enabled for the account.

Step 2: Ask an administrator to enable ActivTrak MCP and approve the platforms

MCP access is deliberately admin-gated. An account administrator enables ActivTrak MCP and selects which AI platforms are allowed, and administrators can later review usage and export an audit trail for compliance review. Agree on the scope before enabling anything: which platforms (Claude, ChatGPT, or both), which teams the rollout covers, and who reviews the audit trail. A sensible first rollout enables one platform for one team, verifies the answers, and only then widens access. If your organization has privacy review requirements, this is the step where that review happens, not after users have already connected their AI accounts.

Step 3: Connect Claude or ChatGPT as a remote MCP connector

ActivTrak MCP is a remote, vendor-hosted capability, not a program you install on your own machine, so there is no local server to download and no command line install step. Follow the connection details in your Early Access onboarding to add ActivTrak as a remote MCP connector in Claude or ChatGPT, then sign in with your ActivTrak account when prompted. Signing in matters for more than authentication: the connection inherits your ActivTrak permissions, so the AI sees exactly the teams and data your account is allowed to see. After connecting, confirm the connector shows as active in your AI app before asking anything. If the connector appears but questions return no ActivTrak data, the usual cause is that Step 2 was skipped, and an administrator still needs to enable MCP or approve that platform.

For the general mechanics of connecting a remote server to Claude, our guide to connecting an MCP server to Claude Desktop explains the client-side concepts, even though ActivTrak’s own onboarding supplies the ActivTrak-specific connection details. The same pattern applies to other enterprise connectors, as shown in our Workday Claude MCP connector tutorial.

Step 4: Ask your first capacity question with explicit filters

Start with a question you can already answer from the dashboard, so you can judge the AI output. A good first prompt names the team, the date range, and the metric in one sentence. For example: “For the support team, for the last four weeks, how did productive time and application usage change week by week, and did ticket resolution slow down in the same weeks?” A second useful question tests the cross-system promise: “Compare the sales team’s work patterns with pipeline outcomes for last quarter, and identify the habits associated with stronger performance.” Always include the date range and the team in the prompt. An AI answer without stated filters is not verifiable, and an unverifiable answer should never reach a staffing decision.

Step 5: Verify every answer against the dashboard before acting on it

This is the step that separates a useful rollout from a risky one. Open the matching ActivTrak dashboard view with the same date range and the same team or group, and compare the headline numbers. They should match, because ActivTrak MCP carries your permissions and privacy controls into the analysis and the underlying metrics are the same. If they match, you have a verified workflow and can move to broader questions, such as comparing AI adoption with changes in capacity and business outcomes. If they do not match, check the filters first: a different week boundary or an unfiltered “all teams” view explains most mismatches. Never forward an AI answer to leadership until at least one number in it has been confirmed in the dashboard.

Step 6: Review usage and export the audit trail

After the first week of use, have the administrator review MCP usage and export the audit trail for compliance review. Check which platforms were used, which accounts connected, and whether any team outside the rollout scope connected. This review is also the moment to decide whether to widen access to a second platform or a second team. Keep the rollout log simple: date enabled, platforms approved, teams covered, and any mismatches found during verification. That log is what turns an Early Access experiment into a governed production habit.

Troubleshooting

The MCP option does not appear in my account

This almost always means the account is not eligible or not yet enrolled in Early Access. Confirm the plan is Workforce Management or Productivity Optimization, confirm you are signed in as an administrator, and confirm with ActivTrak support that Early Access is enabled for the account. Users on other plans, and non-admin users, will not see the option, and that is by design rather than a bug.

The connector is active but questions return no ActivTrak data

Check two things in order. First, confirm an administrator has enabled ActivTrak MCP and approved the specific platform you connected, since enablement and platform approval are separate controls. Second, confirm you signed in with an ActivTrak account that has data for the team you are asking about. An account with no deployed agents, or with visibility limited to a different group, will correctly return nothing for teams it cannot see.

The AI answer disagrees with the dashboard

Do not assume either side is right until the filters match. Re-ask the question with the exact date range, team, and grouping used in the dashboard, and ask the AI to state the filters it applied. Common culprits are week boundaries (Monday versus Sunday start), an “all teams” view versus one team, and productive time versus total time. If the filters match and the numbers still differ, trust the dashboard, save both outputs, and report the mismatch to your administrator so it can be raised with ActivTrak while the feature is still in Early Access.

My team uses Copilot or Gemini, and there is no way to connect

There is no supported workaround today. ActivTrak has announced Copilot and Gemini support for later releases, but Early Access covers Claude and ChatGPT only. Trying to route ActivTrak data into an unsupported platform through unofficial bridges would bypass the admin controls and audit trail that make this integration safe to use, so wait for the official release and keep the dashboard workflow in the meantime.

FAQ

The FAQ entries in the front matter above answer the questions readers ask most: what ActivTrak MCP is, which platforms and plans it supports, whether the AI sees more than the dashboard shows, why answers sometimes differ, and how auditing works. One point deserves emphasis here because it shapes every other answer: ActivTrak MCP is an Early Access capability, and its availability, supported platforms, and onboarding details can change. Treat this tutorial as accurate as of November 2026, re-check the announcement before a large rollout, and keep the verification habit in Step 5 even after the feature leaves Early Access.

Conclusion

ActivTrak MCP is a small setup with a large payoff, provided it is rolled out in the right order: confirm eligibility, let an administrator enable it and approve the platforms, connect Claude or ChatGPT, ask one verifiable question, and check the answer against the dashboard. The teams that benefit most are the ones that stop exporting data for every cross-system question and instead ask the question where the answer already lives, with the same permissions and the same audit trail as the dashboard. Start with one team and one capacity question this week, verify the first answers by hand, and widen access only when verification passes. For your next step, read our MCP server tutorial to understand the protocol underneath, or our Workday Claude MCP connector tutorial to see how another enterprise system connects to Claude the same way.

Frequently asked questions