How Jamf reached 89% active usage in eight weeks with Claude Enterprise in AWS Marketplace

Routine work crowding out what matters

As Jamf grew globally across service desk, infrastructure, and client platform engineering functions, skilled employees were spending significant hours on documentation, communications drafting, information synthesis, and policy navigation. That work was necessary, but it was leaving less time for the analytical depth and creative judgment the same roles required at their best. Rather than letting AI adoption happen organically and inconsistently across departments, Jamf's leadership looked for a way to build a deliberate, governed program from day one.

"Not all work is equal," said Melissa Dunham, Senior Director of Information Technology at Jamf. "Some work requires your full attention, judgment, and creativity. Some just fills your day. Our AI program exists to help every employee find that difference." The program was built around one central question, she added: "What would our employees work on if the routine took care of itself?"

Claude Enterprise as a day-one tool, not an engineering project

Jamf evaluated several enterprise AI platforms on model quality, data privacy, administrative controls, cost, and ease of adoption. Building on an API would have required engineering resources and produced a tool primarily accessible to technical users. Claude Enterprise provided a secure, immediately deployable interface that any employee could use on day one, with enterprise-grade controls that met Jamf's SOC2 and ISO-27001 compliance posture. Jamf chose to procure Claude Enterprise via AWS Marketplace to ensure accelerated procurement cycles and consolidated billing, enabling teams to access Claude Enterprise capabilities sooner.

For the programmatic half of the strategy, Jamf chose Claude on Amazon Bedrock as the API layer for embedded automation. Amazon Bedrock is a platform for building generative AI applications and agents at production scale. “Bedrock fits naturally into our infrastructure and procurement frameworks: No new vendor relationships, no new procurement overhead, and it maps cleanly to our existing cloud security and compliance controls like SOC 2 and ISO-27001,” Dunham said.

The architecture is split by design: Claude Enterprise handles broad employee knowledge work through a governed interface, and Claude on Amazon Bedrock handles developer-controlled API access for embedded workflows that don't need a human at every step. Usage-based pricing also suits engineering-driven workloads where per-seat licenses don't make economic sense.

Claude stood out for the quality and nuance of its responses on the tasks Jamf cared about most: complex analytical work, long-form drafting, and situations requiring careful reasoning. The context window also proved significant for use cases involving large documents and technical content.

Rather than treating governance as a gate that slows adoption to a crawl, Jamf built a model designed to be self-sorting. Developed by Nick Benyo, Principal Solutions Architect on Jamf's Enterprise AI and Automation team, the "One Framework, Three Roads" approach applies one consistent set of principles around data security, acceptable use, and documentation, but provides three distinct paths based on use case complexity.

The Express lane covers approved tools like Claude Enterprise with no special approval required. The Toll Road handles configured use cases (custom Skills, MCP connectors) that need documentation and department-level review. The Off-Road path covers fully custom API implementations requiring architecture and security sign-off. An HR partner drafting job descriptions picks up the tool and goes. A software engineer embedding Claude into a workflow knows they need architecture sign-off. Nobody has to ask which process applies.

Jamf's internal campaign, "Work on What Matters," carries the same intentionality through to its language. "We deliberately avoid words like 'efficiency' and 'optimize,' which feel corporate and cold, in favor of language like 'focus on what only you can do' and 'reclaim your time,'" Dunham explained.

The result: 285 documented use cases across every department without compliance exposure or ungoverned sprawl.

From individual productivity to organizational infrastructure

Claude has become embedded across all 16 departments as working infrastructure. HR leads with 21 implemented use cases with a pipeline of 55 more. In practice, automated new hire data eliminates five years of manual copy and pasting from another software tool. A monthly policy review Skill flags compliance deadlines across regions before they're missed. The team built an HR virtual agent deflecting Tier 0 inquiries before they hit the service desk. Non-technical employees created many of the HR use cases, a proof point for the program's broader democratization thesis.

Marketing's story is about execution velocity: 62.5% of submitted use cases are in production, one of the highest conversion rates in the org. Competitive intelligence, SEO optimization, personalized social content at scale, and AI-powered live chat are all live — compressing the cycle from insight to execution across the entire content operation. Perhaps most telling: Enterprise Transformation, a 15-person team of strategic program managers, has Skills creators representing 60% of its licensed users. The people whose job is to drive organizational change have made building with Claude part of how they work.

One example captures what changes when AI tools become broadly accessible. Jamf's AI program team used Claude to build an interactive department scorecard, processing data from Okta, the company's learning management system, usage reports, and Jira into a self-contained HTML dashboard. The alternative would have been an engineering ticket and an estimated two to four weeks of development. Instead, it was built in roughly eight hours and is now reviewed weekly by all 16 department leaders. The scorecard has driven specific decisions: identifying a department with low access despite strong training completion, surfacing a training gap that led to a gamification initiative, and helping redistribute licenses to employees who needed them.

The same kind of time compression shows up on routine drafting, not just bespoke tool-building. "When Jamf changed its hardware refresh cycle policy, the full communications package was drafted, reviewed, and finalized in about four hours," Dunham said. "Previously, that scope would have required two to three business days. The output quality was high enough that the review cycle dropped from three rounds to one."

The adoption pattern has reached employees who were initially skeptical. Jamf's campaign uses a four-persona framework to meet employees at their readiness level; for skeptics, the message is deliberately low-stakes: "You don't have to overhaul anything. Just try automating one thing you'd rather not do." The most common trigger for conversion is communications drafting. Employees who worried Claude would produce generic output changed their position when they saw a colleague use it to produce a strong first draft they could refine in minutes instead of hours. Heavy document loads are a close second: when employees with compliance reviews or policy research saw Claude process a 40-page document in seconds, the conversation shifted quickly.

The pattern extends to building Claude Cowork, enabled across all Claude Enterprise licenses at Jamf since early 2026, which gives non-technical employees a way to build interactive, reusable tools without engineering involvement. "Bespoke dashboard building for a specific question was impossible before," Dunham noted. Jamf's Facilities Director used Claude to create a dashboard that revealed office patterns at each company location, providing context for data-driven facilities decisions. She built it on her own with no technical assistance, the kind of one-off analytical tool that would never have cleared an engineering backlog.

Demand has outpaced supply. The first 1,000 Claude Enterprise licenses were assigned within four weeks. A waitlist of more than 200 employees built up in days. Jamf is now expanding to 2,000 licenses, already past its 2026 goal of 60% active usage across the organization. API-based implementations are underway through Claude on Amazon Bedrock for more deeply embedded use cases, and the 285 use case library serves as a pipeline for identifying which workflows merit deeper integration.

"It's a completely different mindset we're trying to cultivate," Dunham said. "It's not just 'go use a new tool.' It's thinking differently about how you work, and what your time is actually for." The company plans to close the year by asking each team a simple question: what did you free up, and what did you do with it?