Shannon Denton is Co-Founder and CEO of Wripple, an on-demand talent platform helping companies build and manage flexible marketing teams. Previously Global CEO of Razorfish, Shannon has spent his career at the intersection of technology, marketing, and organizational transformation. Today, he works with marketing leaders navigating the impact of AI on talent, teams, and operating models, with a focus on building more agile, high-performing organizations.
Selected Deliverables
No items found.
About this blog series: AI is changing what marketers can do and how marketing teams get work done. In this series, we explore the evolution of the AI-enabled marketing team, from giving every marketer an AI teammate to building connected, agentic workflows. The goal: human-led, AI-powered teams that deliver better work, faster.
This article explores Stage 3 of our four-stage approach to AI-enabled marketing teams. For the full framework, start with The Marketing Unicorn Is Still Human.
Stage 1 and Stage 2 are about making people more capable. Stage 3 is about changing the shape of the team itself and it's the stage most organizations skip straight past on their way to chasing agents.
I think that's a mistake, because this is the stage that changes the economics.
The idea
Before an assignment starts, you decide how much of the execution AI carries.
Not which tools people use. How the team is built.
Three modes:
Mostly human: 80/20. More human specialists, AI used selectively. Best for high-judgment, sensitive or genuinely original work.
Human + AI: 50/50. Humans supported by AI-enabled talent and tools. Most marketing and creative assignments belong here.
Mostly AI: 20/80. A small human team directing and reviewing AI-heavy execution. Best for high-volume, fast-turn, lower-complexity work.
The percentages are a shorthand for how much of the execution AI carries, not a timesheet. The point is that each mode implies a different team: different number of people, different seniority mix, different cost, different turnaround.
Humans direct in every mode
This needs saying clearly, because the labels invite the wrong reading.
There is no mode where the work runs itself. A human sets the objective, decides what good looks like, and is accountable for what ships. That's true at 80/20 and it's true at 20/80.
What changes across the three modes is how much of the execution AI carries. Not who's directing.
It's not about the type of deliverable
This is the part that surprises people, and it's the most useful thing in the model.
The instinct is to sort by format: strategy is human, production is AI. That's wrong, and it leads to a rigid, unhelpful policy.
Take video. All three modes apply to video:
A brand film runs mostly human. High stakes, long shelf life, genuinely original.
A product explainer runs human + AI. Real craft required, but a known form.
Fifty social cut-downs run mostly AI. High volume, fast turn, and the tenth one doesn't need a director's judgment.
Same capability. Three different team shapes. The mode follows the stakes, not the format.
That's why "we use AI for production" is a weak policy and "we set the mode per assignment" is a strong one.
What you decide it from
Four inputs, in the up-front planning:
The objective - what this work has to achieve.
The type of work - how well-defined the form is.
The speed required - genuinely, not aspirationally.
The judgment and collaboration the work demands - which is the one people underweight.
Note that budget isn't on that list. Budget is an output of this decision, not an input to it. That's the whole benefit.
The payoff
You set cost and speed before the work starts.
Right now, most organizations discover their cost and timeline at the end, after the scope has moved three times. Predefining the mode turns that into a decision made deliberately, in the open, with tradeoffs visible to everyone, including the client or the internal stakeholder.
It also gives you something to negotiate with. "We can do this at mostly-AI in ten days, or human + AI in four weeks" is a real conversation. "It'll take what it takes" is not.
And it makes your talent model legible. A mostly-AI assignment needs two senior directors and a reviewer, not eight specialists. A mostly-human assignment needs the eight. Knowing which you're staffing before you staff it is the difference between a margin and a surprise.
The move worth making this quarter
Take one recurring project type. Define what all three modes look like for it, including team composition, hours, cost, turnaround, and what you'd expect to give up in each.
Then actually run them, over six to eight months, and compare the results. Not a pilot with one sample. Enough reps to see the pattern.
That experiment will teach you more about your AI economics than any vendor evaluation, because it's measured on your work, your standards and your clients.
To learn more about any or all of these solutions, contact your Wripple Client Lead, or request a demo.
About this blog series: AI is changing what marketers can do and how marketing teams get work done. In this series, we explore the evolution of the AI-enabled marketing team, from giving every marketer an AI teammate to building connected, agentic workflows. The goal: human-led, AI-powered teams that deliver better work, faster.
This article explores Stage 3 of our four-stage approach to AI-enabled marketing teams. For the full framework, start with The Marketing Unicorn Is Still Human.
Stage 1 and Stage 2 are about making people more capable. Stage 3 is about changing the shape of the team itself and it's the stage most organizations skip straight past on their way to chasing agents.
I think that's a mistake, because this is the stage that changes the economics.
The idea
Before an assignment starts, you decide how much of the execution AI carries.
Not which tools people use. How the team is built.
Three modes:
Mostly human: 80/20. More human specialists, AI used selectively. Best for high-judgment, sensitive or genuinely original work.
Human + AI: 50/50. Humans supported by AI-enabled talent and tools. Most marketing and creative assignments belong here.
Mostly AI: 20/80. A small human team directing and reviewing AI-heavy execution. Best for high-volume, fast-turn, lower-complexity work.
The percentages are a shorthand for how much of the execution AI carries, not a timesheet. The point is that each mode implies a different team: different number of people, different seniority mix, different cost, different turnaround.
Humans direct in every mode
This needs saying clearly, because the labels invite the wrong reading.
There is no mode where the work runs itself. A human sets the objective, decides what good looks like, and is accountable for what ships. That's true at 80/20 and it's true at 20/80.
What changes across the three modes is how much of the execution AI carries. Not who's directing.
It's not about the type of deliverable
This is the part that surprises people, and it's the most useful thing in the model.
The instinct is to sort by format: strategy is human, production is AI. That's wrong, and it leads to a rigid, unhelpful policy.
Take video. All three modes apply to video:
A brand film runs mostly human. High stakes, long shelf life, genuinely original.
A product explainer runs human + AI. Real craft required, but a known form.
Fifty social cut-downs run mostly AI. High volume, fast turn, and the tenth one doesn't need a director's judgment.
Same capability. Three different team shapes. The mode follows the stakes, not the format.
That's why "we use AI for production" is a weak policy and "we set the mode per assignment" is a strong one.
What you decide it from
Four inputs, in the up-front planning:
The objective - what this work has to achieve.
The type of work - how well-defined the form is.
The speed required - genuinely, not aspirationally.
The judgment and collaboration the work demands - which is the one people underweight.
Note that budget isn't on that list. Budget is an output of this decision, not an input to it. That's the whole benefit.
The payoff
You set cost and speed before the work starts.
Right now, most organizations discover their cost and timeline at the end, after the scope has moved three times. Predefining the mode turns that into a decision made deliberately, in the open, with tradeoffs visible to everyone, including the client or the internal stakeholder.
It also gives you something to negotiate with. "We can do this at mostly-AI in ten days, or human + AI in four weeks" is a real conversation. "It'll take what it takes" is not.
And it makes your talent model legible. A mostly-AI assignment needs two senior directors and a reviewer, not eight specialists. A mostly-human assignment needs the eight. Knowing which you're staffing before you staff it is the difference between a margin and a surprise.
The move worth making this quarter
Take one recurring project type. Define what all three modes look like for it, including team composition, hours, cost, turnaround, and what you'd expect to give up in each.
Then actually run them, over six to eight months, and compare the results. Not a pilot with one sample. Enough reps to see the pattern.
That experiment will teach you more about your AI economics than any vendor evaluation, because it's measured on your work, your standards and your clients.
Shannon Denton is Co-Founder and CEO of Wripple, an on-demand talent platform helping companies build and manage flexible marketing teams. Previously Global CEO of Razorfish, Shannon has spent his career at the intersection of technology, marketing, and organizational transformation. Today, he works with marketing leaders navigating the impact of AI on talent, teams, and operating models, with a focus on building more agile, high-performing organizations.
Building an AI-enabled marketing team? Wripple connects you with vetted freelancers, contractors, and micro-agencies to help you build the skills and capabilities your team needs.
.png)
