Chapter 4: Acceleration vs Expansion
The two ways AI creates real value inside agencies.
Most agencies first experience AI as speed.
Meeting notes appear instantly. First drafts take minutes. Research gets compressed. Rollout gets less painful. The half-day jobs become twenty-minute edits.
Useful. Very useful.
But speed is only half the opportunity.
AI creates value in agencies in two ways:
1. Acceleration
Doing what you already do faster, with less friction.
2. Expansion
Doing things you couldn’t realistically do before.
The agencies making proper progress are building both.
Acceleration removes the drag
Acceleration is where most agencies get their first wins.
It clears the repetitive work, messy admin and awkward middle stages that slow teams down.
Creative teams can use AI to explore territories, test angles, structure scripts, pressure-check ideas and shape stronger first passes.
Strategists can use it to synthesise transcripts, scan categories, cluster themes, interrogate research and build sharper narratives.
Design and production teams can use it for resizing, formatting, visual exploration, consistency checks and rollout.
Client service and project teams can use it to turn messy meeting notes into actions, draft timelines, flag risks and turn internal ramble into client-ready language.
None of this is especially glamorous.
That’s why it matters.
Agencies lose ridiculous amounts of time to friction. Briefing friction. Deck friction. Versioning friction. Feedback friction. “Where the hell did we save that?” friction.
The big glamorous moments rarely kill the team.
The tiny repeated pain points do.
Acceleration removes those.
Done well, it improves speed, margin, consistency, responsiveness and team energy.
More importantly, it gives people time back for better thinking and better craft.
That is the point.
Not more stuff.
Better stuff.
The acceleration trap
Acceleration becomes dangerous when speed becomes the whole prize.
That is how agencies end up with more output, less distinction and a team quietly turned into a content vending machine.
You can already see this everywhere.
More assets.
More variants.
More social posts.
More decks.
More “good enough.”
And less work anyone remembers.
This is the road to AIverage.
Technically competent. Emotionally forgettable.
AI raises the floor of output very quickly. It does not automatically raise the ceiling.
Good agencies use acceleration to create space:
more time to think
more routes to explore
more craft in the finish
more rigour in the strategy
more confidence before presenting
Weak agencies use acceleration to make more mediocre stuff faster.
Congratulations, you’ve invented a very expensive treadmill.
Expansion increases what the agency can do
Expansion is where AI becomes much more interesting.
This is where the agency’s creative and commercial range starts to stretch.
Small teams can now explore, prototype, test and build things that previously needed bigger budgets, larger teams or specialist technical support.
There are two useful levels.
Expansion 1: better exploration
The first level is creative range.
Teams can now explore more territories, visualise routes earlier, test tones, simulate audience reactions and pressure-check ideas before they become expensive.
That changes the quality of development.
You are no longer walking into a meeting with three routes and a slightly sweaty hope that one lands.
You can arrive with better explored thinking.
That does not make senior creatives or strategists less important.
It makes them more important.
Someone still has to decide:
what is strong
what is obvious
what is generic
what is strategically useful
what should be killed quickly
AI creates more possibilities.
Humans decide which ones matter.
Expansion 2: AI-native ideas
The second level is where AI becomes part of the idea itself.
A campaign can behave like a tool, a service, a conversation or a living system.
Think:
a drinks brand with a cocktail engine
a travel brand with adaptive itineraries
a retailer with a style concierge
an event experience that responds to audience behaviour
a brand world that changes based on context
a content system that adapts without losing its soul
This moves agencies beyond making communications.
It lets them build useful, interactive, adaptive brand experiences.
That is a more defensible place to play.
Anyone can make more content now.
Fewer agencies can design a system that creates value for the brand and the audience.
This opens up better commercial models:
prototypes
utilities
licensing
retained optimisation
AI-enabled brand systems
client-facing tools
ongoing advisory
Far more interesting than squeezing another five banners out of the same production budget.
The sweet spot
The strongest agencies will combine acceleration and expansion.
Acceleration protects margin.
Expansion creates differentiation.
One without the other creates problems.
High acceleration with low expansion becomes a content factory.
Efficient. Busy. Forgettable.
High expansion with low acceleration creates chaos.
Great ideas. Messy delivery. Burnt-out team.
Low acceleration and low expansion is the old model with worse odds.
The sweet spot is a faster, sharper, more capable agency with better systems and stronger human judgement.
That is where this gets powerful.
Taste becomes the multiplier
The more AI expands what is possible, the more taste matters.
Infinite options are useless without judgement.
Taste is knowing what to remove.
Taste is spotting the route with tension.
Taste is recognising when something feels generic.
Taste is choosing the idea with emotional truth.
Taste is protecting simplicity when the machine keeps offering more.
This is the bit agencies need to take seriously.
AI can help you explore.
It cannot care for you.
It cannot understand the politics in the room, the twitchy client, the cultural moment, or the odd little human truth that makes an idea land.
That is still the job.
The agencies with the strongest taste, judgement and strategic clarity will get more from AI than the agencies with the longest tool list.
No contest.
The practical move
Most agencies do not need a giant AI transformation programme to get started.
They need a few focused moves:
accelerate two painful workflows
choose one expansion bet
define the quality gates
train people by role
reduce tool chaos
evolve pricing before speed eats margin
That is enough to create momentum.
Start where AI removes genuine friction.
Then find one place where it can create something genuinely new.
Keep humans in charge of judgement, taste and final quality.
Keep the systems simple enough that people actually use them.
Acceleration helps the agency move faster.
Expansion helps the agency move differently.
The agencies that build both become harder to compete with.