Generative AI is quietly changing what companies pay agencies for. One in five CMOs say it's already reduced their need for outside help. But there's one thing AI can't do for you, and it decides whether you end up more independent or just dependent in a new way.
Gartner's 2025 CMO Spend Survey found 22% of CMOs say generative AI has reduced their reliance on external agencies for creativity and strategy, with gains in time efficiency (49%) and cost efficiency (40%). AI is strong at drafts, variants, research support and routine reporting. It can't replace judgment, business context, compliance checks, relationships or accountability. The real win is in-house: your own team using AI, with prompts and accounts in your name.
- 22% of CMOs say GenAI has reduced their reliance on agencies for creativity and strategy (Gartner, 2025).
- CMOs report GenAI gains in time efficiency (49%), cost efficiency (40%) and capacity (27%); only 1% say it isn't a priority.
- AI handles drafts, variants, research support and reporting; people provide judgment, context, compliance and accountability.
- NMC now prohibits AI-generated promotional campaigns by doctors and hospitals and requires AI content to be labelled.
- If the agency holds your prompts and AI accounts, AI has changed your dependency, not reduced it.
What did Gartner find?
In Gartner's 2025 CMO Spend Survey of 402 marketing leaders, 22% said generative AI has enabled them to rely less on external agencies for creativity and strategy (Gartner, 2025). That's roughly one in five, and it came in the same survey where 39% said they plan to cut agency budgets overall.
CMOs also described where generative AI pays off:
What CMOs say generative AI has delivered
Just 1% said generative AI isn't currently a priority. The direction is clear: work that used to be briefed out is increasingly being done, or at least started, in-house with AI.
What AI does well in marketing
- First drafts: posts, ad copy variations, emails, landing page sections and FAQs.
- Volume and versions: resizing, rewriting for different channels and audiences, translating, and creating variants for testing.
- Research support: summarising documents, organising notes, and pulling patterns out of reviews or survey responses.
- Routine reporting: collating numbers and drafting plain-language summaries.
- Ideation: generating twenty angles quickly, so people can pick the two worth developing.
These are exactly the tasks agencies have traditionally billed for by the hour. It's no surprise CMOs are bringing them in-house.
What AI can't replace
Judgment
AI produces plausible options. Knowing which idea is right for your brand, your audience and this moment, and which is merely competent, is still a human skill.
Context
Your customers, your history, what's been tried and why it failed, which claims you can prove: none of that is in a general-purpose model. It lives in the people who've worked on your business.
Accuracy and compliance
In regulated sectors, an AI-written claim can break the rules. India's National Medical Commission now prohibits AI-generated promotional campaigns for commercial interest by doctors and hospitals, requires AI-made content to be labelled, and bans AI-manipulated patient images, voices or testimonials (NMC, Oct 2026). Education, real estate and legal marketing have their own rules too. Someone has to check.
Relationships
Partners, media, communities and customers still respond to people. AI can draft the message; it can't build the trust.
Accountability
When results dip, someone has to own the diagnosis and the fix. McKinsey found the companies getting the most from AI are more likely to have defined how people validate AI outputs, and three times as likely to have senior leaders who own AI initiatives (McKinsey, 2025).
AI vs human: who does what
| Task | AI | Person |
|---|---|---|
| Campaign strategy | Suggests options, summarises research | Decides, based on business goals and context |
| Ad copy | Drafts variations at speed | Chooses, edits, checks claims and compliance |
| Design | Generates concepts and resizes | Sets direction, keeps the brand consistent |
| Reporting | Collates data, flags anomalies | Interprets, decides what to change |
| Customer replies | Drafts and routes routine messages | Handles complex, sensitive and high-value conversations |
| Regulated claims | Can draft only | Must approve every time |
What this means for agency relationships
If AI does more of the production work, the value of any marketing partner shifts from making things to judgment and capability. For founders, that raises a sharper question: who in your business has that judgment?
If the answer is "the agency", AI hasn't reduced your dependency. It's just changed its shape. Now the agency holds the prompts, the workflows and the AI accounts, as well as the know-how. Gartner's Ewan McIntyre put it simply: "AI is not a shortcut around marketing capability" (Gartner, June 2026).
That's why we think the real win from generative AI is in-house: your own team, using AI to do more, with context only they have. McKinsey's research on AI at work found employees are ready, and that their top request is formal training, which 48% said would increase how much they use AI (McKinsey, Jan 2025).
How to bring AI work in-house, safely
- Start with production tasks agencies bill for: drafts, variants, resizing and reporting.
- Keep accounts and prompts in your name. AI tools, prompt libraries and automations should live in your company's accounts.
- Write the rules down. What AI can do alone, what needs review, and what it must never do, especially for regulated claims.
- Train the team and give them time to learn.
- Use partners for what's genuinely specialist, on your terms, with your team in charge.
A college pays an agency to produce 40 social posts a month. With AI and a trained in-house content lead, the college produces the same posts itself in a fraction of the time, using real campus photos and faculty quotes the agency never had. It keeps the agency only for its two big admission-season campaigns.
The agency bill falls. The content gets better, because it's made by someone who walks the campus every day.
For the bigger picture on agency spend, read why 39% of CMOs are cutting agency budgets; on readiness, why only 30% of teams can scale AI; and on people, why people still matter most.
AI can write the first draft. It can't know your business. Make sure someone on your team does.
How Ad Asthra helps you bring AI work in-house
We help your team use AI to do more, with the capability and accounts staying in your business.
Frequently asked questions
Is AI replacing marketing agencies?
Partly. Gartner's 2025 CMO survey found 22% of CMOs say generative AI has reduced their reliance on external agencies for creativity and strategy. Production tasks are most affected; judgment, context and accountability still need people.
What can AI do well in marketing?
First drafts, copy variations, resizing and versioning content, translation, research summaries, routine reporting and generating lots of ideas quickly.
What can't AI replace in marketing?
Judgment about what's right for your brand, context about your customers and history, accuracy and compliance checks, relationships with people, and accountability for results.
Can hospitals use AI for marketing in India?
With strict limits. NMC's October 2026 guidelines prohibit AI-generated promotional campaigns for commercial interest by doctors and hospitals, require AI-made content to carry a source mark, and ban AI-manipulated patient images, voices or testimonials.
Should I let my agency handle AI for us?
Be careful. If the agency holds the prompts, automations and AI accounts, the capability leaves when they do. Keep AI tools and workflows in your company's accounts.
How do I bring AI marketing work in-house?
Start with production tasks, keep AI accounts and prompts in your name, write rules for what AI can and can't do, train your team, and use partners only for genuinely specialist work.
How does Ad Asthra help?
Through BOT we build and hand over an AI-assisted in-house team; through BOM we keep upskilling it; and our Advisory & Governance service reviews what your agencies do so you can decide what to bring in-house.
Sources
- Gartner 2025 CMO Spend Survey Reveals Marketing Budgets Have Flatlined at 7.7% of Overall Company Revenue, Gartner, 12 May 2025
- Gartner Marketing Survey Finds Awareness and Conversion Account for 62.6% of Total Media Spend, Gartner, 8 June 2026
- The state of AI in 2025: Agents, innovation, and transformation, McKinsey & Company, 2025
- Superagency in the workplace: Empowering people to unlock AI's full potential at work, McKinsey & Company, 28 January 2025
- Public Notice: Guidelines on Ethical Advertising and Public Communication by Hospitals/Medical Institutions and Registered Medical Practitioners, National Medical Commission, 6 October 2026
Last updated
Want to work out what AI could take off your team's plate, and what it shouldn't?
Message us on WhatsApp or book a call