CBTS turns itself into an AI testbed as Claude Enterprise rollout delivers rapid ROI

Many organisations have spent the past two years experimenting with generative artificial intelligence, often through small-scale pilots that demonstrate promise but fail to achieve enterprise-wide impact. The challenge has been particularly acute for mid-market firms, which frequently lack the financial resources and specialist expertise available to global enterprises.

Against this backdrop, North American technology services company CBTS, which operates in Canada, has published the results of its own large-scale AI deployment, revealing lessons learned from rolling out Claude Enterprise from Anthropic across its 2,300-person workforce. According to the company, the programme reached return on investment in under three months while recording no major security incidents to date, a finding highlighted in the company’s newly released blueprint for what it terms “100% agentic delivery”.

The initiative is notable because CBTS acted as its own “Client Zero”, using internal operations as a proving ground before extending the approach to customers. Rather than limiting AI use to isolated functions, the company introduced Claude across research, drafting, technical engineering activities and cross-functional collaboration projects, allowing it to evaluate both the benefits and limitations of enterprise-scale deployment.

Moving beyond AI experimentation

One of the recurring themes in enterprise AI adoption has been the gap between experimentation and operational transformation. Many organisations have demonstrated that generative AI can increase productivity in specific tasks such as drafting content, searching information or generating software code. However, integrating those capabilities across a business while maintaining governance, security and compliance standards has proven more difficult.

CBTS says its deployment sought to address precisely that challenge. Employees used different Claude-based tools for a range of activities, including research and documentation through Claude.ai, engineering and software-related tasks through Claude Code, and collaboration through Anthropic’s Cowork platform.

The company argues that this broad rollout enabled it to identify where AI generated measurable value and where human oversight remained critical. Such distinctions are becoming increasingly important as businesses move from experimentation to implementation. While generative AI can accelerate routine tasks and support decision-making, organisations still need experienced staff to review outputs, validate conclusions and manage areas involving risk or regulatory obligations. According to the CBTS announcement, the experience helped clarify the governance structures, workforce changes and infrastructure requirements needed for responsible scaling across larger organisations.

Perhaps the most significant aspect of the programme is the emphasis placed on governance and cybersecurity. Board-level concerns about data protection, intellectual property leakage and regulatory compliance have often slowed AI adoption. Recent surveys from organisations including Forrester Research have found that security and trust remain among the principal barriers preventing companies from moving AI systems into production environments.

CBTS states that security and governance were treated as foundational requirements from the beginning of the deployment. The company says the same framework now underpins its commercial AI services offering, known as Forge AI. In the announcement, CBTS Chief Information Security Officer Chris DeBrunner argued that many chief information security officers face a common dilemma: how to leverage AI capabilities without introducing unacceptable levels of risk. The company’s response has been to establish governance mechanisms that encompass vendor evaluation, threat testing and operational oversight before AI agents enter production environments.

Such concerns are unlikely to disappear. As AI systems gain access to enterprise data, workflows and operational processes, the potential consequences of poor governance increase. The CBTS experience illustrates how organisations are increasingly viewing AI deployment as both a technology project and a risk management exercise.

The rise of agentic AI

The blueprint also reflects growing industry interest in so-called “agentic AI”. Unlike traditional generative AI systems that respond to individual prompts, agentic systems are designed to undertake more complex, multi-step tasks with varying degrees of autonomy. An AI agent might gather information from multiple systems, analyse data, make recommendations and initiate actions as part of a broader workflow.

Analysts increasingly see agentic systems as the next stage in enterprise AI evolution. CBTS references research from Forrester’s 2026 State of Agentic AI report, which argues that major business benefits emerge when organisations redesign workflows, roles and operating models around agentic capabilities rather than simply inserting AI into existing processes. This perspective aligns with wider industry thinking. Technology companies including Microsoft, Google, Anthropic and OpenAI have all invested heavily in platforms intended to support agent-based workflows, reflecting expectations that organisations will seek increasingly autonomous digital assistants capable of undertaking more sophisticated work.

However, such developments also place greater emphasis on monitoring, accountability and transparency. The more independently an AI system operates, the more important it becomes to understand how decisions are made and when human intervention is required.

A key theme of the CBTS announcement is that mid-market organisations often face different constraints from large enterprises. CBTS defines its target market as organisations with annual revenues between approximately $300 million and $3 billion. These companies frequently manage complex operations and technology environments but may lack the dedicated AI teams, extensive experimentation budgets and specialist governance functions available to larger corporations.

As a result, many have struggled to translate AI enthusiasm into scalable programmes. CBTS Chief Executive Officer Kristin Russell argues that the objective is to provide mid-market companies with access to expertise, governance frameworks and operational knowledge that were previously concentrated among the largest enterprises. Whether this approach proves broadly transferable remains to be seen. AI deployments are heavily influenced by organisational culture, data quality, workforce readiness and sector-specific requirements. A strategy that succeeds in one environment may require adaptation elsewhere.

Nevertheless, the publication of operational outcomes from a real-world deployment provides useful evidence in a technology landscape often characterised by ambitious projections rather than demonstrated results.

Lessons for enterprise leaders

The broader significance of the CBTS experience may lie less in the specific technology platform and more in the implementation approach. The company’s findings suggest that successful AI adoption requires attention to multiple dimensions simultaneously: technology selection, cybersecurity, governance, employee training, workflow redesign and executive oversight. Focusing exclusively on the underlying model or software platform is unlikely to deliver sustainable transformation.

The concept of serving as a “Client Zero” is also instructive. By deploying AI throughout its own operations first, CBTS generated practical insights into organisational challenges that cannot always be anticipated in theory. As organisations move from pilot studies to enterprise-wide deployment, those lessons may prove just as important as the capabilities of the AI systems themselves.