Flagship case study
Customer-facing Forward Deployed Engineering for a watsonx-powered generative AI solution progressing from pilot towards production use.
Interactions
5,000+
Engagement supported
$x+
Recognition
Featured by the Financial Times
Confidentiality note
Selected implementation details have been omitted to respect customer confidentiality.
Role and scope
Solly contributed within a wider IBM and customer team. His approved contribution areas include customer-facing technical discovery, core frontend and backend AI components, retrieval and model integration, iterative development and productionisation considerations.
Users and stakeholders
Discovery conversations, workflow understanding and iterative feedback.
Product interface
Frontend experience engineered around real tasks rather than a model demo.
AI and integration layer
Retrieval, orchestration, model integration and production-minded backend components.
Mitsubishi Motors Canada worked with IBM on an intelligent companion experience powered by watsonx. Solly contributed within a wider IBM and customer team, with the project becoming the strongest example of his customer-facing AI product delivery work to date.
The goal was to build a useful generative AI experience around real user needs rather than a demo in search of a problem. That meant balancing user-facing product quality with the realities of enterprise delivery, integration and stakeholder alignment.
As with many applied AI engagements, requirements evolved through discovery, feedback and validation. Solly worked in an environment where the exact shape of the solution needed to be clarified through collaboration, iteration and technical judgement rather than handed over as a complete specification.
Solly supported customer-facing technical discovery, built core frontend and backend AI components, worked on retrieval and model integration, and helped iterate the system based on stakeholder and user feedback. He did this as part of a wider team and does not claim sole authorship of the full product.
The solution combined user-facing interface work with backend AI orchestration, retrieval and integration layers. Solly engineered core full-stack components and helped shape how the experience connected model outputs to practical user workflows.
The work moved through customer conversations, iterative implementation, stakeholder review and production-minded refinement. Solly’s contribution included translating feedback into product and engineering changes while keeping both technical feasibility and user value in view.
The solution progressed from pilot towards production use, recorded more than 5,000 interactions and supported a commercial engagement worth more than $x. It also received external recognition through coverage by the Financial Times.
The project reinforced the value of combining discovery, product judgement and full-stack execution in one role. In ambiguous AI work, the hard part is rarely just connecting a model endpoint, it is deciding what should be built, how it fits the user journey, and how to iterate it responsibly inside enterprise constraints.
What I owned
Decision summary