Most organizations aren’t debating whether to adopt AI. Their teams already have. The real question is what that adoption actually produces, and whether anyone can verify the answer.
Unstructured AI use doesn’t generate zero value. In fact, it generates negative value. Incompatible practices across teams. Quality gaps that accumulate from one project to the next. As a result, senior engineers spend time fixing ungoverned output instead of focusing on work that requires their judgment. Behind all of it, blind spots persist around data confidentiality, regulatory compliance, and production code quality.
These aren’t theoretical risks. They exist in every organization using AI without a defined framework often without anyone having consciously decided to accept them.
That’s why we treat this as a governance question, and not only as a technology question.
CBTW is a Global Tech Solutions company. In practice, that means we cover the full technology value chain: understanding business workflows, building applications, infrastructure, security, data exploitation, and demonstrating measurable impact. That depth of integration doesn’t leave room for surface-level familiarity with AI tools our teams use daily.
So we started by applying that standard to ourselves.
Starting with our own teams
Before recommending anything to our clients, we made a deliberate choice to understand, test, and develop genuine proficiency with these tools internally. Because advice only holds up when it comes from experience, not assumption.
Every quarter, an internal group analyses and stress-tests available solutions against real production environments, maturity, actual use cases, and integration into existing workflows. Claude Code stood out for one reason that matters in delivery context: it produces directly usable results. For development teams working on complex codebases with real quality and maintainability requirements, that’s the distinction that counts.
We chose to certify most of our team with Anthropic. The difference is straightforward: a publisher-issued certification validates proficiency against the publisher’s own standards. It’s an external validation we cannot issue ourselves.
That said, Anthropic certification is not our only training path. We complement it with additional learning resources tailored more directly to the specific roles and disciplines of our Software Engineering teams. The goal is to combine verifiable external credibility with practical, role-relevant depth.
In a market where every technology partner claims AI expertise, that combination certified foundation plus targeted upskilling is how we back it up.
What the certification covers
The Claude Code certification validates operational proficiency in real development contexts. Specifically, it covers effective use in a terminal environment, integration into existing workflows, and understanding of the model’s capabilities and limits. It also covers practical application across core Software Engineering use cases: code generation, review, refactoring, documentation, and quality assurance.
The bar is set by Anthropic. Passing means a developer has demonstrated a level of proficiency that the publisher itself recognizes.
For our teams, that represents a genuine investment of time and effort. We give them the space and resources to prepare and earn them properly. The certification then becomes a personal professional asset, something that belongs to the developer himself. We are also integrating it into career progression across the group, which signals something beyond a one-time initiative.
A program in progress
We set ourselves a target of 320 certified developers by the end of May 2026. At the time of publication, 65% of them have already completed the certification, and the program is still ongoing.
An initiative announced only after the fact isn’t a transformation. It’s a press release. Showing the program as it runs, with its milestones and its trajectory, is more useful information for anyone evaluating a technology partner.
To clarify: the program currently covers all development teams across Europe within the Software Engineering service line, with plans to extend it to our other service lines and countries. The goal is straightforward, to make Claude Code proficiency a standard, not a specialty.
What this means for clients
For our clients, this translates into something concrete. When a CBTW team works on their codebase, the developers involved have been evaluated against an external standard, not an internal one we designed ourselves. That distinction matters. It means the proficiency has been verified by a third party with a direct interest in setting the bar correctly: the publisher of the tool being used.
In practice, clients get a team that already knows how to use these tools effectively. They also benefit from ongoing training tailored to specific roles. Concretely, that means faster integration of AI into development workflows, more consistent output quality, and a team that understands both the capabilities and the limits of the technology they work with.
Furthermore, working alongside developers who use these tools as a natural part of how they work is one of the most direct ways for a client organization to start moving in the same direction.
What this signals about how we work
Certifying 320 developers is also a concrete commitment to our team. In a market where AI proficiency is increasingly a differentiator, recognition carries weight beyond CBTW.
Conclusion
A technology investment must produce measurable impact, otherwise it doesn’t hold over time. AI is no different. AI is already part of the projects we deliver and the proposals we write. It integrates as a component of delivery, not as a separate promise. Certifying our teams with Anthropic is one step in building something verifiable around that external, documented, and grounded in what we actually deliver.
If your software engineering projects deserve better than self-declared AI expertise, reach out: odellamaggiore@cbtw.tech
