The person who can design AI best is YOU, the domain expert.

Rather than asking “How much work can AI reduce?”,
ask “What new value can AI add to this work?”

Good AI Results Begin with “Work Context,” Not Technology

When we first encountered generative AI in 2023, we tended to think first of complex technologies and code. However, Palette believes that before deciding what to ask AI, we first need to be able to explain whose problem we are trying to solve, what that problem is, and what should be considered a good outcome.

|GOAL Solve customer or business problem

|Quality Criteria the dividing line between good and bad outcomes

|Basis and Exceptions Product Information · Brand Guidelines · Prohibited Expressions

|Review Authority The scope that a person must review and approve

Applied AI Is an Approach That Transforms Technology into Real Business Value

In this article, Applied AI does not refer to a specific model or product. Rather, it refers to a practical approach in which artificial intelligence is applied to real problems and workflows to create business outcomes and new possibilities.

Palette defines it as: “AI that is not technology for technology’s sake, but is applied to people’s work to create new value.”

For example, generative images and videos can reduce constraints related to location, weather, travel, and budget, expanding the range of creative possibilities. The key is not to replace on-site production altogether, but to test, on a smaller scale and more quickly, ideas that were difficult to realize through conventional production methods, while people review brand strategy, factual accuracy, copyright, and disclosure requirements. This is not simply about reducing our workload or replacing existing work by doing it faster and more cheaply.

We first consider the various roles and decision-making authorities involved in the work in order to understand its context and determine whose problem we need to solve and what that problem is.

That Is Why Domain Experts Should Design AI Themselves.

Tacit knowledge—such as unwritten rules and exceptions learned through experience—is best understood by domain experts. This is why Human-in-the-loop design is essential, allowing experts to set goals, define evaluation criteria, and review results. No-code AI agents allow even non-developers to incorporate this context conversationally and iteratively refine the AI to fit their own workflows.

Expand expertise keep judgment, but broaden the scope of execution.

Redesign work reduce repetition and focus on problem definition and
decision-making.

Validate possibilities start small with ideas previously delayed due to cost and technical barriers.

Experience the Unexperienced

The people who use AI best are not those who know every technology, but those who know what matters in their work and how far they want to transform it. Palette creates the easiest environment where that domain expertise meets AI.