A leading member of our digital leadership community became Group CIO of a technology company just as the sector was turned on its head. The experienced CIO has been here before, having led technology in the insurance sector during Brexit, as well as a data company, and logistics at times of major growth and consolidation. This experience in navigating a business and its technology function during a time of major change has proved useful, as a leader for a software provider.
The CIO and their team have all had to adapt rapidly to the impact and opportunity of AI-led software development. Speaking to our digital leadership community, they revealed how to help technologists and staff benefit from the system change AI will create.
We have anonymized the CIO in this article to protect the community.
Starting with an initial 100 and then rolled out across the business, the CIO invested in retraining the firm’s staff, in particular its software engineers, to be skilled AI users. Using lessons from the initial 100 retrained staff, they created a mass of material, a schedule of hackathons and lunch-and-learn events to bring everyone’s skill levels up, and to reduce fear. It is not as easy as it sounds, as they say:
Not everybody wants to come on the journey, and there were some difficult board meetings about those people that don’t; in addition, AI was not applicable to every product we have. But we had to be consistent.
Staff were given six to nine months to join the reskilling initiative. The CIO says there was a fear factor towards AI, which is only natural, especially during a time of a cost-of-living crisis and economic uncertainty. Some of the fear was a lack of knowledge. As the reskilling took place, it iteratively improved. In addition, a training engine was developed so that software developers could look at what they had created using AI, and then work with senior engineers to improve and learn.
Of late, we have heard a great deal about token costs, with big league vendor bosses stepping into the debate. Initially, our CIO community member had to set aside token cost worries to get people comfortable and learning AI. They add:
To start with, we were all writing poor prompts, but now we are seeing some gamification as to what tasks and token efficiency can be achieved.
That is creating a sense of shared learning across the software engineers and is something the CIO advocates:
We have been demonstrating, demonstrating, and demonstrating, not just how they build with AI, but how they use the tools to get the best value and solve problems. We did it every week for six months, and it still continues.
Some engineers did leave the organization as they believed AI was going to be a “lever” to get them out of the business, but our CIO says:
It wasn’t. We went public on this being a value creation opportunity, as we want to build more products, and we publicly stated that if you are on the bus, then your job is safe.
AI-led SDLC
AI is transforming the software development lifecycle (SDLC) in enterprises and technology providers. Traditionally, the CIO explains, business analysts would write a specification that the engineers take and build from. Today, this organization is using Claude from Anthropic to write the specification, and then code is developed using both Claude and Devin from Cognition.
The two different AI development tools are used because Devin integrates well with Microsoft Azure DevOps, and Claude integrates well with GitHub, the CIO says. The majority (70%) of developers are using Devin, and 30% are using Claude. They add:
The difference between the two is the back-end repo (repository) that you are trying to integrate with, and the level of parallel engineering that is required.
I think we will remain a split shop so that I have some competitive friction and resilience, which is important as they are still both new businesses with constantly changing business models, and they are not making money yet.
But what about the quality of the outputs?
We have written brand new products, and we are blown away. For example, we had a product that was written in Progress that needed modernizing. We ran Claude over it for the analysis and then used Devin to rewrite it. That was initially a project that was going to take 18 months; we did it in four months with 24 employees. We just can’t believe the speed, quality, and capability of the AI-development platforms.
As the developers increasingly use AI to develop software, they are free, the CIO says, to take a more granular view of the SDLC and products; they add:
We look at DORA statistics, cycle times, and we are now building a portal that provides a 360-degree view of engineering velocity, the bugs we are seeing, backlogs, and vulnerabilities to get the balance of costs and value. Costs have gone up, but the cycle times have significantly shrunk.
Meanwhile, an automated load distribution AI tool that had been deployed identified 117 vulnerabilities in 19 products overnight, which it sent to an AI-bug buster that fixed these and sent them to an engineer at 6.30am, who was able to review and send them to production by 9am.
Code pull requests have increased three times, and the CIO expects there to be more automation added to the SDLC; they add:
It has really helped us build new products. We now debate whether to refactor using agentic tools or do we start from scratch? My preference is to start from scratch as it is quick enough.
Security
High-speed business change makes security teams nervous, and the CIO admits that was the case for this firm. They add:
We have a great CISO, and you need to have a really great team around you, as AI is a team effort as we are trying to balance speed, cost, innovation, and security.
Technology development, of course, played a role too:
We built an AI analysis engine that looks across the estate and who is using what. On launch, we found we had 6000 different AI tools across our estate.
That discovery meant the CIO and their peers in cybersecurity had to strike a balance between encouraging the use of AI and having the visibility necessary to protect the business and its customers. They add:
We didn’t want to come down with a hard stick, so we built a portal as an education tool that showed what was inside and outside of our AI policy. That provides staff with an indication of what is approved and what are the alternatives they can use if an AI tool is not approved, it also shows them what to do to get a tool approved. We did this as we realized we had to educate the firm on the sheer volume of AI tools.
Once again, it is that focus on educating everyone in the organization on the opportunities, but also the risks, that way they all get to learn and feel included in the system change AI is creating.
My take
AI is going to be a system change, the way the internet, computing, mechanization, and the industrial <a href="https://bitcomme.com/nitda-tech-revolution-africa-builders-festival-2026/” title=”NITDA-Tech Revolution Africa Builders Festival 2026″>revolution were before them. A people-centric move protects not only our society and communities during the system change, but also, as this CIO’s story reveals, the businesses CIOs work for.
