Most conversations about AI start with efficiency. Take a process that already exists, automate the repetitive parts, and save the team some time.
But automation is only one way to transform the way you work.
You can also use AI to revisit the process itself. Would you still design the work the same way? Could you add something that was never practical before? Could the process produce a new service rather than simply producing the same deliverable faster?
The IDEA Model helps teams think through those possibilities.
How the IDEA Model works
The model has two axes. One is novelty: do you want to replicate or improve something the team already produces, or create something new? The other is how closely the work follows existing human logic: do you want AI to work within the way people do the job today, or use it without those constraints?
None of the four is inherently better. The choice depends on what the team wants to preserve and what it wants to make possible.
Deconstruct
Break down the human logic behind the current work and rewire it.
Imagine
Create products, processes, and systems without human constraints.
Emulate
Faithfully replicate the human process.
Advance
Scale, accelerate, or continuously optimize an existing process.
The four approaches
Emulate
Faithfully replicate the human process.
Emulate sits closest to traditional automation. Use it when the current process works and the team wants AI to perform part of it without changing the sequence, inputs, or decision rules.
Advance
Scale, accelerate, or continuously optimize an existing process.
Advance preserves the existing process or deliverable but uses AI to operate at a scale or speed that would not be practical for a team to sustain manually.
Use Advance when the team knows what it wants to produce but lacks the time to review enough sources, test enough variations, or tailor the deliverable for each audience.
Deconstruct
Break down the human logic behind the current work and rewire it to build something new.
Deconstruct starts with something the team already produces. The team asks what purpose it serves, identifies the information, reasoning, structure, and other components that make it useful, then recombines those parts to create something new.
Use Deconstruct when the current work contains valuable thinking or information, but the finished deliverable is not the only—or best—way to put those components to use.
Imagine
Create products, processes, and systems without human constraints.
Imagine begins with a capability the agency wants rather than a process or deliverable it already has. The team considers what AI makes possible when the design no longer has to mirror the way people perform the work today.
Use Imagine when the agency wants to create something it could not realistically staff, scale, or sustain through a human-run process.
IDEA in action: Competitive intelligence
Start with the work as it exists today. An agency produces a monthly competitive intelligence report. The team reviews a familiar set of publications, competitor websites, social accounts, and analyst sources; records significant developments; compares activity across competitors; and summarizes the implications for the client.
Emulate
Follow the same process
AI checks the usual sources, applies the agency’s established categories, and drafts the report in its current format. The process and output stay the same; AI performs some of the work.
Advance
Expand what the process can handle
AI reviews a much broader set of sources, compares developments across competitors, and tracks patterns over time. The output is still the monthly report, but it draws on more information and analysis than the team could reasonably produce by hand.
Deconstruct
Recombine the components
The team breaks the report into the parts that make it useful: individual developments, source evidence, comparisons, recurring themes, and implications. It can then recombine the relevant components around a specific client decision. Instead of another monthly report, the output might be a planning tool that organizes competitor activity around the client’s priorities, campaigns, or upcoming decisions.
Imagine
Create a new capability
The agency moves beyond the monthly report and creates an always-on opportunity and risk sensing service. AI connects changes in hiring, messaging, product announcements, leadership commentary, media coverage, and customer conversations, then surfaces developments that merit a strategist’s attention. The strategist decides what each signal means and how the agency or client should act.
Start with what you want to achieve
IDEA can start with a problem the team wants to solve or an opportunity it wants to pursue. In either case, define what you want to achieve before choosing an approach.
Reduce the manual work in a process you already trust
EmulateWho or what performs the steps
Scale, accelerate, or continuously improve an existing process
AdvanceHow much the process can handle
Use the components of existing work to create a new output
DeconstructHow those components are combined and what they produce
Build a product, process, or system that does not have to mirror a human-run process
ImagineWhat the agency can create or provide
Quality is still the standard
Efficiency only matters if the work remains good. Across all four approaches, we use the same test:
Is this as good as, or better than, what we produced before AI?
The amount and purpose of human review changes with the approach. In an Emulate workflow, a person may check whether AI followed the established process. In an Imagine build, a person may decide whether the system’s recommendation is strategically sound.
In crisis work, for example, AI can track coverage and conversations across more sources than a person could read in real time. It can detect unusual changes, organize the evidence, and compare the current event with previous ones. People still decide whether to respond, what to say, and how to account for the relationships involved.The IDEA Model helps teams consider the full range of AI applications: preserving a process, extending it, rebuilding it, or moving beyond it. Each leads to a different outcome. The path forward is not to push every idea toward Imagine, but to decide deliberately what AI should change about the work.