AI for
Marketing
Agencies
AI changes the cost of research, production and reporting, which makes time-based agency pricing harder to defend. We help established agencies redesign the delivery workflow, protect client data and measure the work clients still value.
22%
OF CMOs CUTTING AGENCY USE (GARTNER)
~9%
WPP HEADCOUNT CUT IN 2025
$500K
SORRELL'S NEW PRICE FOR A $2.5M AD
AI is changing how agency work is priced
Sir Martin Sorrell has said his teams now deliver for $500,000 in four to six weeks what used to cost $2.5m and four months. He told Digiday the billable hour is dying and he's moving Monks onto subscription-style fees because the maths on time-and-materials doesn't work anymore.
WPP cut headcount by roughly 9% in 2025 and has dropped freelancer use by a quarter over two years. Forrester forecasts another 15% reduction in agency jobs in 2026, after an 8% cut in 2025. Rory Sutherland, Ogilvy UK's vice chairman, has said there is "no connection at all between the hours you spend on a problem and the value of the solution."
If the holdcos can't defend the hourly rate, you can't either. The agencies that come out the other side will sell outcomes instead of timesheets.
YESTERDAY'S AGENCY
- Bills time, not outcomes
- Juniors writing first drafts
- Deck factory, briefing chains, status calls
- Client questions the agency's value beyond ChatGPT
- Margins compressing every renewal
REBUILT AROUND AI
- Priced per output, per asset, per result
- Drafts written by your own agents, on your brand
- Brief, deck, variants and report all generated
- You're the AI capability they're buying
- Margin held because cost of delivery collapsed
Agency work worth automating
Not slide decks about AI. Working software running inside your agency, against your tools, on your brand. The stack we build with creative and media shops.
Brief to deck
An RFP or client brief produces a first draft in your house template, using relevant past work as context. A senior reviews and edits it instead of starting from a blank slide.
Copy on brand
A model fine-tuned on your client's tone-of-voice docs, past campaigns and approved copy. Variants for paid, organic and email in one pass. CAP code checks built in.
Asset factory
Midjourney, Firefly, Runway and ElevenLabs wired into one pipeline with brand guardrails, licence logging and an approval queue. Static, video, voiceover, sized for every placement.
Reporting that writes itself
GA4, Meta, TikTok, LinkedIn, search console, Looker. Pulled, summarised, written up in your account team's voice. Monday morning client emails, drafted before they wake.
Client call to action
Every call transcribed, decisions extracted, actions assigned, contact reports drafted. The bit account managers hate doing, and now don't. Searchable across every account.
How we start
Vu is an agency and operates its own AI products and internal agent workflows. That gives us a direct view of model cost, client approval, quality control and the commercial questions that appear after a demonstration.
The audit identifies one client or internal workflow with a measurable commercial case. A build phase covers its data, approval path and operating cost, with ownership of prompts, configured models and code set out in the proposal.
BOOK AN AGENCY AUDITAgency audit
We look at your timesheets, your retainers and your delivery process. We tell you which of your hours are already commoditised by ChatGPT, which clients are at risk, and the three workflows to automate first. Priced and named up front.
Capture the agency brain
Brand books, tone-of-voice docs, past decks, winning RFPs, client backgrounders. Loaded into one place your AI can read. Without this every agent sounds like ChatGPT. With it, they sound like you.
Launch the first agent
Pick the workflow with the worst margin. Brief to deck. Or weekly client reports. Or asset variants. Live, used by the account team straight away. Hours that used to be billed at junior rates now happen overnight.
Re-price the offer
We work with the commercial lead to define the outputs, service levels and approval work in the retainer. The agency can then price the result and capacity rather than passing model usage straight through.
What agency clients are changing
Three things published this year, none of them predictions. Each one makes the next client conversation harder.
22% of CMOs already cutting back.
Gartner's 2025 CMO Spend Survey, 402 CMOs across North America, the UK and Europe. 22% said GenAI lets them reduce reliance on agencies for creativity and strategy. 39% said they're planning to cut agency budgets outright.
Your clients are behind you.
95% of UK digital advertising businesses use AI. Across UK business as a whole, the number is 16%. Your CMO has read three articles. You're already running the tools. That gap is the most sellable thing you've got.
A 100% gen-AI campaign delivered.
Brandtech reported using Pencil for a generative campaign for French leather house Lancel. It is a useful public example of generated creative moving into paid client work rather than remaining an internal demonstration.
Sources: Gartner 2025 CMO Spend Survey (gartner.com, 12 May 2025); IAB UK "Powering Growth: The UK's Hidden AI Engine" (iabuk.com, May 2026); Brandtech Group / Pencil case study for Lancel, 2024; Digiday on Sir Martin Sorrell and S4 Capital/Monks (March 2026); Forrester "Predictions 2026: Marketing Agencies" (forrester.com).
What agency leaders are saying
"The billable hour is dying."
SIR MARTIN SORRELL, S4 CAPITAL / MONKS, DIGIDAY MAR 2026
"There is no connection at all between the hours you spend on a problem and the value of the solution."
RORY SUTHERLAND, VICE CHAIRMAN OGILVY UK, THE DRUM FEB 2026
"The end of the world isn't nigh, it's AI."
SIR JOHN HEGARTY, BBH FOUNDER, THE DRUM APR 2024
When this is worth discussing
We work best when there is a real operating problem, enough volume to measure and people from the affected teams who can make decisions.
Usually a good fit
- An established UK business, usually with annual revenue above £10m
- A repeated process with a known cost, delay, error rate or capacity problem
- A senior sponsor and a day-to-day owner who understand the work
- Access to the relevant staff, systems, sample records and security requirements
We may point you elsewhere
- A standard product already covers the process well
- The requirement is a one-off small build with no wider operating case
- There is no owner or access to the people and data needed to test the result
- The plan relies on AI making high-impact decisions with nobody responsible for review
Questions the buying team will ask
Won't this just make my juniors redundant?
Some production work is changing quickly, but junior staff still need a route to learn judgement, clients and craft. We design the workflow so automation removes defined production steps while people retain review, client context and development work.
My creatives say AI work is soulless. Are they wrong?
They're right about the bad version. Coca-Cola's 2024 AI Christmas ad got called "soulless" in the press for a reason. We don't replace your creatives' taste. We give them tools that take the drudgery off, so the work they put out has more of their thinking in it, not less.
Doesn't ChatGPT or Copilot already do this?
For a marketer in-house, often yes. For your agency, the point isn't access to a model, it's the system around it. Your brand books, your client tone-of-voice docs, your past decks, your approvals, your asset library, your placement specs, your media plans. None of that lives inside ChatGPT. We build the layer that does.
What about ASA and ICO rules on AI content?
CAP's May 2025 guidance is clear: no blanket requirement to disclose AI use, but the CAP code on misleadingness applies regardless of how content was made. The ICO has said legitimate interest is the only available basis for training generative models on web-scraped personal data, and that vague transparency notices won't pass the balancing test. We build that in, so your client doesn't end up in front of either of them.
We serve EU brands. What about the EU AI Act?
The prohibited-practices and AI-literacy provisions have been in force since 2 February 2025. Article 50 transparency obligations on chatbots and deepfakes apply from 2 August 2026. Prohibited-practice fines reach €35m or 7% of global turnover; transparency breaches up to €15m or 3%. Your client can have duties as a deployer. We include the disclosure logic, training and audit trail in the build.
How do we re-price retainers without losing the client?
Describe the service in outputs, review points and response times, then decide how any efficiency affects margin and price. We can help the commercial lead model the retainer and explain which work remains human-led.
How much does it cost?
The audit is fixed-fee. The first agent is priced before we start building, scoped against the audit. After that it's per workflow, fixed per phase. We tell you the number up front. You own everything we build.
Who owns the prompts, fine-tunes and code?
You do. That includes the source, prompts, fine-tune weights and evaluations. Host it with us or take it in-house. We don't lock agencies into our hosting.
Talk to us about your agency workflow
Bring representative retainers, delivery workflows and recent margin pressure. We will identify the work AI can reduce, the judgement clients still buy and a first internal system worth testing.