The promise of AI is appealing to a CEO with a long growth agenda and a CTO with an already full roadmap. More ideas can be tested. Repetitive work can be reduced. Smaller teams may be able to do more. Yet every new AI tool can also create another integration, security review and support obligation.
I ran into this tension repeatedly as a growth and transformation consultant. We could show where better use of data, a redesigned customer journey or a new digital service might create value. The organisation still needed people to specify, build, test, integrate, operate and improve it. The strategy and the capacity to deliver it lived on different sides of the table.
Why I built Kinsugi
Kinsugi grew from that frustration. Gain helps a business and technical team turn an objective into a project outline and retain the decisions made along the way. KADE, the underlying development platform, uses small, reusable software modules with defined jobs. Engineers can inspect the components, connect them to existing systems and see the automated checks applied during development.
Suppose a retailer wants an assistant that helps a customer choose a product. A useful pilot needs more than a persuasive conversation. It needs current stock, price rules, product suitability, permission to act and a route to a person when the system is unsure. A business leader defines the commercial objective and the limits. Engineers own the connections and technical review. The assistant should operate within those boundaries.
The checks are part of the engineering evidence. A validator passing means a particular rule passed at a particular point; it does not certify that a finished application is secure, compliant or commercially effective. Integration testing, security assessment, user testing and human judgment remain essential.
That distinction matters now. McKinsey's 2026 survey found that 32% of respondents said their organisations had decided against buying at least one software product or feature because they could build the functionality internally with AI coding tools. A build decision still needs a proper view of maintenance, ownership and total cost. The same survey found that AI operating costs constrained usage for about one in five respondents.
A different working relationship
Our model is simple: we do it for you, we do it with you, then you do it for yourself. Consulting can help identify the right opportunity and its economics. Kinsugi engineers can help with the first solution. Your own people learn the method, take part in reviews and gain the ability to extend and operate what is built. The handover is designed into the engagement rather than left for the final week.
For a CFO, the test is whether the solution changes revenue, cost, quality or cycle time after full operating costs. For a CTO, it is whether the software can be understood, changed, secured and supported by the team that inherits it. Both need a voice before a pilot begins.
If you want to test a specific business problem, register for a free Kinsugi account and outline it with Gain. For a company pilot, we can then agree a sponsor, success measure, technical scope and review points together.
Which capability would you want your team to own by the end of a pilot rather than keep buying from a supplier?