As B2B marketers, we’re deep into the next phase of AI. The focus has shifted from experimentation to activation – from tools to transformation. At our latest Bright Marketing Leaders Dinner, we brought together a brilliant group of marketing and business leaders to explore what it really takes to embed generative and agentic AI in marketing, and, crucially, to prove the value.  

We were also joined by Max Gabriel CEO at Augmented AI who provided an expert perspective on where Agentic AI is going and how marketers can start to operationalise it so that agents take care of the repetitive stuff and marketers can shift focus to what really matters: strategy, creativity and meaningful engagement.  

Held at Noble Rot in Mayfair, the evening centred on three practical questions: 

1. Where is AI delivering value right now? 
2. How are we measuring its impact? 
3. What’s standing in the way of scaling it? 

Here’s what came out of the discussion – grounded insight from leaders in the thick of it. One thing was clear: agentic AI is still in its infancy, with most teams only experimenting with application-based AI or single-purpose agents in isolated use cases, but there’s a shared ambition to understand, operationalise, and scale its role across marketing as quickly as possible.  

Agentic AI: From automation to intelligent collaboration 

There was strong consensus around the table: we’re past the point of automation for automation’s sake. The conversation has shifted to how AI, specifically agentic AI, can act as a collaborator, not just a tool. These agents don’t just complete tasks; they make decisions, respond to context, and work in concert with other systems and tools.  

The promise? Marketers can move away from repetitive, manual work and focus on higher-value thinking. That shift changes the skill profile too – less emphasis on specialism, more on critical thinking, data literacy, stakeholder management, and the ability to interrogate and improve AI-driven outputs. 

A standout theme was the concept of ‘agent crews’, sets of AI agents working in coordination across a campaign or marketing operation. Imagine one agent curating content, another analysing performance, and another adjusting messaging based on real-time engagement data. Each has a role, they talk to each other, and they evolve as the work unfolds. Everyone agreed this was an area to develop but no-one was close to this concept yet.   

But that future raises important questions: How do we manage and maintain these agents? Who’s responsible for prompt optimisation, drift correction, or performance tuning? There was talk of creating an “agent ops” function – a new capability that acts like a team coach and tech lead rolled into one, overseeing agent quality and relevance just like you would with any team member.  

It also became clear that as AI agents become more embedded, we can’t afford to import poor processes or bad habits into automated workflows. This moment requires unlearning – rethinking how we collaborate with machines and investing in training that teaches teams not just how to prompt, but how to think with AI. 

Some teams are already going AI-first in their planning – starting with the question: “Where can AI add value here?” They’re building test-and-learn loops directly into campaign workflows, from strategy through to execution, with a view to scale what works. 

Leaders have a critical role here. It’s not just about giving access to tools – it’s about creating the conditions where teams feel confident to explore, experiment and work with AI as a partner. It also means shifting the narrative: using AI isn’t ‘cheating’ – it’s a core skill for modern marketers. 

Already, we’re seeing teams spin up synthetic focus groups to test and refine propositions, fast-track market research, and explore gaps in buyer journeys. Done well, it’s a route to faster go-to-market and more predictable performance. 

Defining value: rethinking AI’s impact beyond cost savings  

When it comes to AI in marketing, too much of the conversation is still focused on cost-cutting and headcount reduction. But this mindset is both premature and limiting. AI, especially in its agentic form, isn’t yet capable of running operations end-to-end. In fact, according to Gartner1, agentic AI can currently handle only around 15% of day-to-day work decisions without human intervention, with the rest still requiring input for strategy, interpretation, refinement, escalation and governance.  

As the technology matures, that percentage will grow – but for now, marketers need to focus less on replacement, and more on redesigning work to get the best from both people and agents. So rather than asking, “How much cost can we cut?”, the better question is, “How do we redesign work so that humans and agents operate better together to drive grow and revenue?” The reality is we’ll have to find ways to achieve both.  

This reframing is crucial. If we don’t rethink the human-agent relationship, we risk simply layering new tools onto old workflows – increasing complexity, not value. Many attendees raised concerns about creating more operational overhead if AI isn’t integrated in a deliberate, structured way.  

That’s where agile marketing models are already proving their worth – offering clear workflows, transparent metrics, and sprint-based feedback loops that make it easier to embed AI into day-to-day activity, test and iterate, and track time-to-market gains and efficiency improvements. 

At the same time, there’s a need to focus on where AI can drive growth, not just save costs. Agentic and application-based AI are already: 

• Accelerating content creation and testing, shortening lead times 

• Improving campaign segmentation and personalisation, increasing relevance 

• Generating insights from synthetic audiences, helping shape better propositions 

The shift in search behaviour is another wake-up call. Traditional SEO performance is declining as more traffic comes via chat-based search and conversational interfaces. This opens up new questions: how do we optimise for AI-driven discovery? What does success look like when your audience is no longer clicking through, but getting summaries and recommendations from agents? 

While advertising via AI interfaces is still in its infancy, leaders are watching closely. It’s early days, but brand influence may become a critical lever – how your brand is presented, mentioned, or prioritised by generative models could impact awareness and consideration. Brand metrics like share of synthetic voice, AI-indexed reputation, or visibility in curated responses may soon be standard indicators.  

Meanwhile, we’re also seeing a move by larger corporates to recentralise marketing operations – pulling back from distributed field teams to create more standardised, scalable models that are AI-ready. This is partly driven by cost pressures, but also a recognition that data, tools, and processes need to be integrated and governed centrally to support AI-led execution.  

But here’s the tension: pressure from the boardroom to cut costs fast, or slash headcount prematurely risks destabilising current marketing models. These agents are still nascent. We’re still working out what good looks like. There’s broad agreement that marketing needs to reinvent itself – and agentic AI will play a key role – but this requires investment in capability, operating model, and culture. 

That means shifting how we measure success – not just by cost savings, but by the value created at every step of the workflow. That includes speed, accuracy, creativity, personalisation, and business outcomes like pipeline, engagement quality, and revenue contribution. 

Scaling AI means scaling capability – not just tech 

The ambition to scale AI was clear – but so were the blockers. Two major themes emerged: capability and confidence.  

Many teams still lack the skills or structures to use AI in a way that feels safe, effective and measurable. From prompt engineering to critical thinking alongside AI, the capability gap is real. As one guest put it: “Some people are brilliant prompt engineers. Others need more structure, frameworks, examples – and permission to experiment.”  

But capability isn’t just about individual skills – it’s about building confidence at the team level, and fostering a culture where experimentation is expected, not exceptional. That means leaders must shift their focus from performance oversight to enabling their teams to work in new ways, encouraging group learning and creating space to try, test and refine how AI is used. 

There was also healthy debate around organisational readiness. It’s one thing to introduce new tools – it’s another to adapt your operating model to make them work. Agile marketing was seen as a crucial enabler here – not just as a methodology, but as a mindset. Agile ways of working support AI adoption through sprint-based testing, prioritised backlogs, and visibility of outcomes, giving teams the structure to embed AI iteratively, without overhauling everything at once. 

And then there’s governance. With agents acting semi-independently, how do we ensure risk is managed and brand integrity maintained? There’s growing recognition that AI governance needs to be proactive, not reactive. Leaders shared early examples of building guardrails and escalation paths, defining what AI agents can and can’t do, and how to intervene when things go off track. 

Underpinning all of this is a bigger shift in how we think about work: we must start asking higher-level questions, such as “What’s the best resource – human, agent, or hybrid – to deliver this outcome?” That means rethinking workflows, roles, and decision-making frameworks – and being clear about where AI adds the most value, and how to measure it.  

Scaling AI isn’t just a technical rollout – it’s a transformation programme. The most successful marketing teams will be those that invest in capability, reimagine how their teams operate, and embed agentic AI in a way that’s strategic, not just opportunistic.  

Final word: the winners will be those who change how they work 

If there was one shared takeaway from the night, it was this: the value of AI in marketing doesn’t come from the tools – it comes from how we work and the data that underpins them.  

The marketers who will win in this new era, aren’t those with the most tools. They’re the ones who rewire their ways of working to be more responsive, experimental, data-led and human-centred. They’re enabling their teams, building the right culture, and proving impact where it matters most.  

This dinner wasn’t just a great conversation. It was a clear signal of where the smartest B2B teams are heading next. 

Interested in joining our next dinner? 

We’re curating the next conversation now. Drop us a message to register interest and join a growing community of marketing leaders navigating the AI shift with purpose.