Leading with governance, building a blueprint customers can follow

Demand outpaces the toolset

Headquartered in Switzerland, SoftwareOne is a global software and cloud solutions provider operating in more than 70 countries with 13,000 employees worldwide. Its relationship with Microsoft dates back 25+ years, and its internal rollout of Microsoft 365 Copilot, which began in 2023, became one of the company's most visible AI success stories with an adoption rate that outpaced industry averages by a wide margin.

That experience set a precedent. When Microsoft began integrating Anthropic's Claude models into the Microsoft 365 Copilot ecosystem, SoftwareOne faced a familiar question with a new twist: not just whether to adopt a new AI capability, but how to do so in a timely and responsible manner that would match the advice it offers clients weighing the same choice.

By early 2026, requests for Claude access were arriving from across the business. David Tannheimer, SoftwareOne's Head of AI Internal Advisory: IT & Solutions, watched the pattern build for months.

People had noticed a gap. Copilot delivered strong compliance and security credentials, while Claude was widely recognized for its capabilities in writing, analysis and coding work.

Interest in advanced AI capabilities continued to grow across the business, highlighting the importance of making new technologies available within a secure and governed environment. SoftwareOne recognized this opportunity early and took a proactive approach, working to extend its existing governance framework to support additional AI capabilities while maintaining high standards for security, privacy, and compliance.

TIMELINE Customer Zero journey at a glance
When	What happened
Jan–Feb 2026	Demand for Claude builds across developer teams, presales, and the NORAM region.
Early Feb 2026	Enablement approached as a business and governance decision, rather than solely a technology initiative.
Late Feb 2026	Escalated through established governance channels for review and evaluation.
Early Mar 2026	Endorsed through established organizational processes.
Mid Mar 2026	Approach agreed: proceed with a structured and governed rollout.
March 2026	Claude enabled inside Microsoft 365 Copilot.
Late March 2026	Microsoft launches Copilot Cowork, built at first exclusively on Anthropic's models — SoftwareOne is ready from day one.
Summer 2026	A 50-user Claude Enterprise pilot launches alongside developers and the NORAM team, with 300+ more colleagues already in the queue. A NORAM-led comparison finds Claude Cowork and Copilot Cowork at parity on quality.
Fall 2026	An eight-week Cowork evaluation with Microsoft begins, aimed at sharpening ROI evidence. SoftwareOne pursues Anthropic Preferred Partner status, scaling its Enterprise pilot og wider AI tooling strategy in tandem.

 

Governance first, always

Enabling access to Anthropic's models was never just a technology decision. It also required careful consideration of governance, compliance and operational requirements to ensure organizational readiness.

SoftwareOne's governance, security and compliance stakeholders were involved from the outset, building on the approach established for earlier enterprise AI deployments.

As interest in Claude continued to grow, the focus was therefore on understanding how it could be introduced responsibly within the organization's existing governance framework. A comprehensive review considered a range of operational, security and compliance factors, alongside the potential value the technology could bring to employees. 

Rather than treating governance as a gate that would delay innovation, the teams worked in parallel. Controlled testing and evaluation helped stakeholders better understand the technology, while ongoing review activities ensured it could be introduced in a way that aligned with organizational standards and requirements.

 

A timely decision

The outcome was a pragmatic path forward. By building on existing governance foundations and enterprise protections, SoftwareOne was able to make Claude available within its broader AI ecosystem while maintaining the same focus on security, compliance and responsible use that had guided previous AI initiatives.

Anthropic Claude went live inside SoftwareOne's Microsoft 365 Copilot environment during March 2026. At the time, nobody knew quite how significant that date would turn out to be. Just weeks later during the same month, Microsoft released Copilot Cowork, its new agentic capability for multi-step work across Microsoft 365.

In its first iteration, Cowork ran exclusively on Anthropic's models.  Organizations that hadn't yet enabled Claude therefore found themselves locked out of testing Cowork altogether. SoftwareOne wasn't one of them.

This overlap wasn't luck so much as persistence. Momentum had been building for months, with teams exploring how emerging AI capabilities could be introduced in a way that balanced innovation with the organization's governance requirements.

That meant SoftwareOne could start using Cowork from its earliest release, an advantage that fed directly into the next phase of the story.

Cowork becomes the breakout use case

Ask people at SoftwareOne which Anthropic-powered capability changed how they work, and the answer is consistent: Cowork. Where Copilot chat helps with a single question or a quick draft, Cowork takes on defined, multi-step work pulling from files, calendars, emails, and Teams conversations to deliver a finished document, deck, or plan, largely without the back-and-forth iteration earlier AI tools required.

For many users, it was the first time Copilot did what they had hoped it would do since 2023: read the brief, do the work, and then hand back something close to finished.

That impression has since been objectively tested. A comparison exercise run with the NORAM team put Claude Cowork and Copilot Cowork side by side. The result: the two reached parity on quality, giving SoftwareOne confidence that its Microsoft-integrated route to Claude wasn't a compromise on capability and represented the very best combination of AI tools currently available.
Practical benefits were quickly being realized across the organization.

When SoftwareOne's Global Bid Management team needed to rebrand a client-facing deck running to hundreds of slides, a Claude-based skill, built specifically for the job and running through the deck in smaller batches, reproduced it at close to full accuracy. 

Today, Anthropic's models – mainly Claude Sonnet and Opus 5 – sit behind Copilot Studio, Word, Excel, PowerPoint, Copilot chat, and Cowork, with model selection left either to the user or to Microsoft's automatic routing. Alongside that, SoftwareOne is also running a separate, direct Claude Enterprise pilot with 50 users, working closely with developers and the NORAM team.

The program has already attracted eager interest among colleagues more widely.

Strengthening proof

There are good reasons for this level of demand: colleagues across the organization continue to demonstrate just how powerful Anthropic models can be; they now want to maximize the impact and improvement on their work through broader access to Anthropic models.

In one example, a SAM consultant in SoftwareOne’s ITAM practice built a Claude-based skill to handle the hardest part of software licensing renewal analysis.  This is the time-consuming manual work that starts once SAM tools have done the first 90 percent of licensing tasks. Tested against a consultant doing the same review by hand, this skill delivered impressive accuracy. This was a difference worth more than $100,000 a year for one customer alone and saving more than 330 hours of consulting time. The consultant started with Windows licensing and is already extending the same approach to other vendors.

SoftwareOne is now working to connect examples like these to company-wide measurement systems, the same way it advises clients to: one measure pilot at a time.