For the modern B2B marketer, the AI revolution can often feel less like a streamlined future and more like a high-stakes construction experiment.

While the market is teeming with tech companies offering task-oriented "marketing agents" designed to automate isolated workflows, specialist marketing solutions providers say the industry is solving the wrong problem. Sandeep Menon, co-founder and CEO of agentic marketing platform Auxia and former vice president of marketing for payments at Google, argues that large brands' core challenge isn't automation, since they already have plenty of that. What's missing, he says, is context and intelligence.

Independent research backs up that diagnosis. Forrester's 2026 predictions describe the industry at an inflection point, where deterministic, rule-based automation no longer defines the frontier and the growth of agentic AI has moved the goalposts from task execution toward contextual reasoning and adaptive decision-making.

Yet Forrester expects fewer than 15% of companies to actually switch on the agentic features already built into their automation suites, an indication that the industry's automation layer is mature, but its capacity to reason with context is not.

The thirteen-week bottleneck

Enterprise marketing has long suffered from fragmentation. In most large organizations, data is trapped in deep, functional silos.

"Nobody in a company has a complete context from a marketing perspective," said Menon. "There is a lot of data, the web team has this data, the app team has some other data... so nobody has the end-to-end context, and that I think is holding back marketing."

Without a unified view of the customer journey, executing a truly integrated campaign in a large enterprise can take up to 13 weeks. By the time a campaign finally launches and performance metrics trickle back, the market has moved, and many of the learnings are already too late to be actionable.

This reliance on rigid, rule-based workflows keeps marketing teams tethered to highly repetitive, manual tasks and prevents them from acting as the strategic, high-impact "10x marketers" their businesses need. Experts say the solution is not to plug more disconnected AI bots into the system, but to implement a unified command center that establishes context across the entire marketing stack.

Off-the-shelf vs. purpose-built

Many generic AI vendors offer off-the-shelf agents that require complex technical "plumbing." Marketers end up managing API connections, navigating unpredictable token costs, and worrying about system compatibility. This forces teams to focus on the technology itself.

This "control plane" framing is becoming a common pitch across the category, even as vendors take different paths to it. Auxia, built by a team of Google veterans, is one specialist provider making that case directly: The company positions its platform as a command layer sitting above a marketing department's existing tools rather than another point solution to manage.

Auxia isn't alone in reaching for that framing, though the approach varies depending on where a vendor is starting from. Adobe has pitched a similar vision through Journey Optimizer, part of its Experience Platform, describing the move from AI as a support tool to AI as an autonomous partner making decisions and orchestrating personalized experiences at scale, but layered onto legacy enterprise stacks. Salesforce is similarly betting on orchestration layers with Agentforce Marketing, folding agentic decisioning into the infrastructure of what was previously pre-AI Marketing Cloud.

Incumbents are adding reasoning and context on top of systems that were originally designed to follow fixed rules, while challengers like Auxia seek to bake in decisioning from the start.

The marketer’s action plan

The consensus among analysts is that the bottleneck isn't the software. Gartner predicts more than 40% of agentic AI projects will be canceled by 2027, citing unclear ROI and weak governance as the leading causes.

Menon puts it more bluntly: "People often think that AI is a technology problem... It is not a technology problem. It is more an organizational problem. If you can't change the way your team works, there's no point in having the latest technology."

That framing points to four steps industry researchers and practitioners increasingly agree on:

1. Stop waiting for pristine data: Start with your existing data sources. AI will help you identify which sources are actually driving performance, allowing you to refine your data collection and instrumentation over time.
2. Avoid AI "pet projects": A Zapier survey of over 800 senior leaders found that 84% of companies have at least one AI pilot that never made it to production and that efforts are often stalled by IT infrastructure, data quality, and system integration issues, not the underlying AI itself. Teams should demand a clear ROI and business outcome before launching any initiative, rather than green-lighting AI projects to satisfy boardroom pressure.
3. Solve the organizational problem: AI implementation is primarily an organizational challenge, not a technological one. If you do not restructure how your team collaborates, breaking down the silos that cause 13-week delays, the most advanced AI in the world will offer little value.
4. Demand model agnosticism: Avoid getting locked into a single "magic" model. A robust enterprise solution should automatically route tasks to the most efficient model, whether that means a high-capacity model or a cheaper, faster local model, freeing marketers from technical micromanagement.

Context and organizational change, not more agents, are what move marketing past basic automation. The teams that fix how they work first are the ones whose AI will be worth having.

This was originally featured in the EMARKETER Daily newsletter. For more marketing insights, statistics, and trends, subscribe here.