Why automating a broken workflow with AI is a trap

There’s an old urban planning legend about paving the cow path. The story goes that when the city of Boston was being laid out, the planners didn’t design new roads, but paved over the wandering, meandering trails that cattle had worn into the dirt over centuries.

The resulting network goes in nonsensical circles, takes twice as long to traverse and confuses everyone who tries to navigate it. Right now, marketing organizations everywhere are doing the same thing with artificial intelligence.

We’re under immense pressure to adopt AI. We’re buying tools, licensing copilots and training teams on prompt engineering. But in our rush to modernize, we are often taking our existing, convoluted, approval-heavy, siloed workflows — think of them as our corporate cow paths — and simply adding AI to them.

The result is bad processes happening quickly. If you apply AI to a workflow riddled with friction, you get chaos faster than efficiency.

As I prepare for my upcoming MarTech Conference session on March 4 about rethinking marketing workflows, I want to offer a different approach. Before you license another tool, you need to audit the path itself.

Avoiding random acts of AI

In my work helping organizations transition to become AI-native enterprises, I see plenty of what I’ve dubbed random acts of AI. This happens when teams adopt tools in isolation. A copywriter uses ChatGPT to draft emails. A designer uses Midjourney for mockups. A data analyst uses a plugin to clean spreadsheets.

On the surface, this looks like progress. But if you look closely at the way work flows through the system, nothing much has actually changed.

The copywriter still has to email the draft to the manager.
The manager still lets it sit in their inbox for two days.
The designer still waits for the approved copy before starting the layout.
The final asset still gets stuck in a compliance review loop.

We have optimized the tasks (writing, designing) but ignored the system. We have paved the cow path, but we’re still walking in circles.

To truly build a hyperadaptive organization (one that can sense, respond and evolve in near real-time, with the help of AI), we need to stop treating AI as a tool for individuals and start treating it as a catalyst for systemic redesign.

In my upcoming book, “Hyperadaptive,” I introduce the concept of the dual engine. This is the idea that you must run two workstreams simultaneously:

Process optimization: Identifying and removing friction from your workflows.
AI integration: Applying technology to the optimized steps.

You can’t do one without the other. If you optimize without AI, you miss exponential gains. If you integrate AI without optimization, you automate waste.

Dig deeper: Implementing AI without a problem is a fast road to failure

Fixing the flow alongside AI

Here’s the practical, four-step methodology we use in our applied AI workshops to fix the workflow before we touch the tools.

Map the friction points

Gather your team in person or virtually and map out a specific marketing process, such as a campaign launch or asset creation. Don’t map what should happen. Map what actually happens.

Look specifically for organizational friction and fidelity loss.

Friction: Where does the process stop? (e.g., Waiting for legal approval, Waiting for data access, Formatting slides manually).
Fidelity loss: Where does information get distorted? (e.g., The creative brief is handed off three times and by the time it reaches the designer, the original strategy is lost).

AI thrives on clean data and clear signals. If your current process relies on hallway conversations and implied approvals, AI will fail. You need to identify these friction points and ask: Is this step necessary, or is it just how we’ve always done it?

Dig deeper: It’s time for AI to join your workflows

Apply the AI automation inversion

Most leaders ask, “Which tasks can AI automate?” I want you to flip that question. If machines handled the routine work, what uniquely human capabilities should we amplify?

When we worked with a marketing team on their newsletter process, they realized their current workflow involved five handoffs among four people. Even with AI tools, the process took four days because of wait states — the time work sat in someone’s queue.

By inverting the problem, they realized they didn’t need a faster writer; they needed an automated workflow. They redesigned the process so an AI agent could detect a new blog post, draft the newsletter snippet, generate the graphic and stage it for review automatically.

The human role shifted from writing and pasting to reviewing and strategizing. The cycle time dropped from four days to one hour. That is the power of fixing the flow.

Move from augmentation to agents

Finally, decide on the level of intervention.

Augmentation is using AI as a tool (e.g., help me write this subject line). This is good for individual tasks.
Automation (agentic AI) is allowing AI to execute a workflow (e.g., when a lead comes in, score it, research the company, draft a personalized email and add it to my draft folder).

The most significant productivity gains in 2026 will come from moving from Augmentation to Agents. But agents require structure. They need clear rules. They require a paved road, not a dirt path.

Dig deeper: How AI agents will reshap

Don’t let busywork derail you

The irony is that we often are too busy to fix our processes. We’re drowning in the very manual work that we claim we want to automate. This is the AI time paradox. We’re too busy bailing water to patch the hole in the boat.

To break this cycle, you have to be willing to stop. You have to carve out the time to map the value stream. You have to be willing to look at a process you’ve used for five years and say, This doesn’t make sense anymore.

When you take the time to fix the workflow first, AI stops being just another tool to manage. It becomes the engine of a new way of working.

Join the conversation

On March 4, I’ll be joining Brianna Miller and Moni Oloyede at The MarTech Conference for the live panel “Say goodbye to busywork by rethinking marketing workflows.”

We’re going to move beyond the theory and dig into the reality of: