8. I’ve been doing a lot of jigsaw puzzles lately.

It’s the first time I’ve picked one up since the peak of the COVID lockdowns. There is something incredibly therapeutic about it. Sorting the chaos, finding the edge pieces, building the frame. It quiets the immediate noise and it allows the mind to start to wander.

Lately, my mind has been wandering to the job market.

I’ve been doing a lot of networking and outreach over the last few weeks, reading through endless job descriptions and agency manifestos. Almost every single one screams the same two buzzwords: "AI-First!" and "AI-Driven!"

Everyone wants an AI-first employee. But here is the multi-million dollar question nobody is actually answering: How are you going to measure them?

If we use AI to eliminate the "$20 tasks", the data entry, the basic reporting, the silo-chasing, we can no longer measure an employee's value by how many widgets they produce or how many hours they bill doing grunt work. If the machine writes the code, builds the baseline media plan, and aggregates the data, what is the human actually responsible for?

Measuring the Architecture, Not the Tasks

To measure an "AI-First" team, you have to stop looking at task completion and start looking at how your talent impacts the core architecture of the business. I think about this down in four pillars:

• Commercial & Growth Engine: Winning and expanding. External revenue generation and market positioning.

• Operational Architecture: Running efficiently. Internal structure, flow, systems, and codifying knowledge for scale.

• Financial & Talent Core: The P&L health, margin realization, and the development of the people.

• Client Success & Value: Translating the first three pillars into actual, undeniable commercial value for the client.

The AI Integration Maturity Model

You can't hire an "AI-First" executive and drop them into a completely manual agency. To measure their impact, you need to understand where your organization's infrastructure currently sits on the Strategic Client Capacity scale:

1. Reactive Foundation (<10% Strategic Client Capacity) Your infrastructure relies on manual effort, siloed data, and tribal knowledge. Processes are functional but inconsistent and slow to scale (e.g., disconnected Sales, Ops, and Finance systems). There is a poor connection between Sales and Media teams. You aren't AI-first; you are just surviving.

2. Standardized Alignment (10% - 25% Strategic Client Capacity) Systems are finally defined, documented, and standardized across functional areas. Data is actively captured, but it still requires manual aggregation for cross-pillar insights. Reactive issues across the organization are identified and planned for, but the heavy lifting is still very human.

3. Predictive Orchestration (25% - 50% Strategic Client Capacity) This is where AI starts paying the bills. All core pillars (Sales, Ops, RevOps) are interconnected and feed a central data layer. The infrastructure allows for automated forecasting, cross-channel optimization, and immediate insight into profit margins and growth friction. Because the machines are humming, Media & Client Services have the bandwidth to deliver value that actually drives renewals.

4. Human in the Loop (50%+ Strategic Client Capacity) The pinnacle. The core infrastructure is primarily AI/LLM-driven, automating high-volume, complex decision-making (e.g., dynamic pricing, staffing allocation, partnership identification). The human executives are no longer managing tasks; they focus solely on strategic design, quality assurance, and solving the creative, non-linear problems the machines can't touch.

The Picture on the Box

If you want to build an "AI-First" organization, your business needs to be ready to measure people on outcomes, not outputs. You measure your leaders on how quickly they can move your agency from a Reactive Foundation to Predictive Orchestration.

It all goes back to the jigsaw puzzle.

You can have the most powerful AI in the world, but you can't just dump 1,000 pieces on a table and expect the picture to assemble itself. You need someone with the experience to find the edge pieces. You need someone to build the framework.

Most importantly, you need a Human in the Loop who actually knows what the picture on the box is supposed to look like.