CIM’s hype-free guide to AI agents for marketers

September 7, 2025|Kim's Blog, Marketing, Statistics and trends|

In August, I attended a CIM webinar Events | CIM on hype-free guide to AI agents for marketers. The presenter was Kerry Harrison. It was a fascinating talk providing a glimpse of the future covering creativity, productivity and automation for marketing. It was possibly the first time I fully grasped both the real, long-term potential of AI in marketing as well as the possible impact of day-to-day life and skills for future marketing professionals. CIM’s hype-free guide to AI agents for marketers.

For me, previous eye-opening events on AI in professional services marketing included:

In February 2024, Alastair Banks presented an excellent webinar (hosted by Grace Williams, co-chair of PM Forum South West and Head of Marketing at Michelmores) on Generative AI – Going Beyond the Content… (pmint.co.uk) – member access only

Lessons from digital marketing webinars (June 2023) – Ben Lee of Bidwells talked about making ChatGPT work for you at a Property Marketing Network Home – Profile (profile-network.com) event

Kerry Harrison – Augmenting human creativity with AI

Kerry Harrison | LinkedIn is a founder and director, AI trainer and speaker, award-winning AI practitioner and course director of sessions on AI copywriting, AI enabled content production and AI agents for marketing productivity.

Her work as an AI curator for the Cheltenham Science Festival took her onto BBC Radio. She is a course director at CIM.

CIM Artificial Intelligence (AI) courses include:

AI in Marketing | CIM
AI For Copywriting – how to use AI copywriting tools to develop new and innovative ways to create content
AI Programmatic Advertising – how to optimise your programmatic advertising strategy and set up effective campaigns
AI to Develop Customer Insights – how to use AI tools to build propensity models to predict customer actions, churn, and value
AI Enabled Content Production – how to use AI tools to generate text, images, animations, and videos
AI Customer Journey Optimisation – how to use AI tools to attract, engage, convert, and retain customers
AI Agents for Marketing Productivity | CIM

There are also AI-related courses at PM Forum – PM Forum – see, for example: Key Insights from the AI in Marketing training by Optix at PM Forum

What is an AI Agent?

Kerry said it took a lot of time to define an AI agent.

AI agents are software systems that use AI to pursue goals and complete tasks on behalf of users. They show reasoning, planning and memory and have a level of autonomy to make decisions, learn and adapt. So key elements:

Goal-driven rather than rule-driven
A level of autonomy to perform tasks
Complex, multi-step actions
Memory – to maintain context, learn, improve or adapt

She provided a helpful metaphor to aid understanding:

Automation – follow a recipe
- Predetermined steps: repetitive rule-based tasks: if this, then do this
AI Automation – a McDonald’s team member
- Follow the rules and clear instructions
AI Agent – Restaurant chef
- Chooses tools available, adapts to meet requests, can remember what has worked

AI trends

Whilst acknowledging it is early days, Kerry shared some statistics showing AI agent trends:

88% of executives say their team or business function plans to increase AI-related budgets in the next 12 months due to agentic AI (PwC 2025) AI agent survey: PwC
67% of surveyed senior executives agreed that AI agents will drastically transform existing roles within the next 12 month (PwC 2025)
AI agents increase 55% of the efficiency and reduce 35% of the costs for companies using systems such as Salesforce
Gartner predicts at least 15% of day-to-day work decisions will be made autonomously through agentic AI by 2028 (up from 0% in 2024)

She also mentioned the Microsoft report: 2025: The year the Frontier Firm is born where it says that every organisation’s AI transformation will look different but likely to follow:

Phase 1 – Human with assistant – Every employee has an AI assistant that helps them work better and faster
Phase 2 – Human agent teams – Agents join teams as “digital colleagues” taking on specific tasks at human direction
Phase 3 – Humans set direction and agents execute business processes and workflows, checking in as needed. Every human becomes an agent boss.

But she warned about the hype:

Gartner predicts over 40% of agentic AI projects will be cancelled by end of 2027
Most agentic Ai projects right now are early stage experience or proof of concepts that are mostly driven by hype and are often misapplied
Many vendors are engaging in “agent washing” – the rebranding of existing products, such as AI assistant, robotic process automation (RBA) and chatbots
In this early stage, Gartner recommends agentic Ai only be pursued where is delivers clear value or ROI

She mentioned that custom GPT applications were being called agents. And showed an M365 Copilot Agent Store. It listed possible agents such as analyst, LinkedIn blog-style post, bold copy, researcher, career coach, prompt coach, newsletter snipper.

Examples of AI Agents

She talked through some simple examples of automation, AI automatoin and automation with AI agents – for example:

Cognitive autonomy scale

She described the cognitive autonomy scale as we move from human-guided to self-directing AI

She talked briefly about democratising Ai development – from non-techies to coded AI agents via No-code AI agent builders. She described democratised AI as being similar to the way that ChatGPT democratises writing.

No-code AI Agent builders mentioned:

Make.com (visual builder – agent capabilities released April 2025)
N8N – visual builder – no-code and developers – self-host option
Relevance.ai (quick conversational build but less visual)
ChatGPT Agent (available within ChatGPT, use prompting, no build)

Risks

She provided a quick overview of some of the risks:

Prompt injection – hidden instructions in content/email could mean fraudulent transactions or leak sensitive data

Data overexposure and leakage – Agent could access broader data than necessary, integrations increase attack surface

Unintended actions/consequences – Agents could delete, send emails or make transactions without proper validation. Hallucinations and bias

Lack of accountability – Traditional legal frameworks can’t quite keep up. Whilst having human-like intentions, the “black box” can make it difficult to trace

To mitigate the risks, we can

Keep a human in the loop especially for critical decisions and high-risk actions
Limit permissions – follow least-privilege principles and connect only necessary apps
Check the terms and conditions of apps you connect with and use trusted sources
Education your teams about accountability and the protocols and practices to follow

AI use cases

Some of the use cases she described:

Support with content strategy – plan, source, analyse and structure your content research
Creating content – generating ideas from guidelines, creating posts, scoring them based on criteria and improving them
Simulating customer journeys – using agents to explore your site and obtain feedback
Lead scoring and nurturing – analyse incoming leads, score them based on criteria and send follow-up emails

She described a ChatGPT Agent (Plus needed) that conducted audience or customer research. She also demonstrated how she obtained 10 content ideas and posts from a brand guide.