Applied AI & Product Development
Applied AI is the point where an AI idea has to work for a real user. I start with the task and the decision that needs support, then work out what type of solution is justified.
Start with the task, not the interface
It is easy to begin with a format such as a chatbot, agent, copilot or dashboard. I prefer to start earlier. What information does the user need? What part of the work is repetitive? Where is judgment required? What should the AI do, and what must remain with a human?
This usually leads to a better product decision because the interface becomes a consequence of the work, not the starting point.
What gets built
The result can be an internal workflow that improves analysis or reporting, an AI-enabled service offered to clients, a reusable method that teams can apply repeatedly, or a custom product built around a specific need. Not every useful AI idea should become software.
Custom development makes more sense when the problem recurs, the workflow is clear, the information structure is stable enough and a consistent user experience creates value that cannot be achieved with an existing tool.
From prototype to regular use
I normally work through the problem and user first, then the required inputs and sources. A prototype is useful for testing whether the model and workflow can produce a useful result before investing in interface or infrastructure. After that, representative cases, domain review and source checks show where the solution fails.
Only then does it make sense to connect the capability to the actual process, roles and controls. Once people use it, their feedback becomes part of the product work.
Public examples from the Curious Ahead AI portfolio
The public AI AHEAD work includes AI-driven capabilities for reputation intelligence, AI visibility, reporting, media training, crisis simulation and decision support. The AI Ahead Reporting Tool and Reputation Scoring Formula received Gold and Bronze respectively at the AI & Data Awards 2025. PRISM, Message Elaborator and Predictive Reputation Simulator received three Silver distinctions at the AI in Marketing Awards 2026.
These are Curious Ahead / V+O portfolio examples developed with the team. They are not products owned personally by me.
Build the capability so it can change
Models and vendors change quickly. I therefore prefer products where the business value sits in the problem definition, domain knowledge, source structure, workflow, evaluation and user experience around the model. That gives the solution a better chance of surviving changes in the underlying technology.