Nearly eight in ten companies have deployed generative AI in some form. Yet roughly the same proportion say it has made no material difference to their bottom line.
This “gen AI paradox” is caused by an imbalance between easy-to-deploy use cases and the more complex, high-value ones.
Most organizations have rolled out employee copilots, but often these only deliver siloed benefits to specific users. Higher-impact vertical- and function-specific use cases rarely make it out of pilot phase – often held back by technical, organizational, data, and cultural barriers.
Part of the problem is that organizations are layering AI on top of existing ways of working, rather than redesigning workflows around how humans and AI should collaborate. That redesign is what unlocks end-to-end transformation – and, ultimately, value.
From assistance to execution: how agentic AI unlocks value
This is where agentic AI comes in.
Agents don’t wait to be asked. They reason, plan, and act across systems – handling multi-step processes without needing human approval for every decision. This shift from “AI as assistant” to “AI as executor” is where real commercial impact begins to show.
One of the clearest areas of value is marketing and sales. Some Fortune 250 companies have reported campaign creation and execution speeds increasing fifteen-fold. It’s more than just automation – it’s a reallocation of work between humans and agents across the campaign process.
As companies introduce agents into the marketing mix, the focus shifts to redesigning workflows end-to-end – across teams, data, systems, and agents – so they operate in unison. With the right structure and discipline, agents can act, decide, and collaborate with human oversight.
That’s when results become tangible: predicting shifts in customer sentiment, triggering outreach before a customer calls, and resolving cases pre-emptively with personalized offers.
For instance, one European insurer re-architected its commercial model around a connected network of agents in just sixteen weeks. As part of the project, AI went from reviewing three percent of sales calls to automatically reviewing 95 percent.
It’s a great example of the internal transformation story: agentic AI as the execution layer that turns gen AI’s promise into operational and revenue impact.
The consumer behavior imperative
But there’s a second shift underway – one that is less within a company’s control. While businesses are still working out how to deploy agents internally, consumers have already started using them externally.
Our recent study across France, Germany, and the UK found that 84 percent of consumers use AI tools in everyday life. What starts with chatbot tools is quickly extending into brand-owned agents and embedded agentic experiences.
Fifty-five percent are using AI to learn about categories and products. Forty-six percent use it to discover new things to buy. Crucially, this is happening before a consumer visits a website, clicks an ad, or interacts directly with a brand.
AI is becoming the primary interface for discovery, comparison, and – increasingly – transaction initiation. The implication is uncomfortable but real: by the time a customer reaches a brand, much of the decision has already been made.
Paid search, creative, and conversion rate optimization all still matter. But they’re adjacent to where decisions are being formed. The question is shifting from “how do we convert customers?” to “how do we stay visible and relevant when the first interaction isn’t with a human at all?”
Building trust in agentic commerce
This is where “agentic commerce” comes into focus.
Agentic commerce refers to a model where AI agents act on behalf of consumers to discover products, evaluate options, make recommendations, and – eventually – execute transactions.
The scale of this shift is significant. By 2030, agentic commerce could orchestrate between $3 trillion and $5 trillion in global transaction volume. What sounds abstract at first is, in practice, a fundamental rewiring of how demand is created, shaped, and captured.
Today, much of this activity sits in the discovery and decision phase. Execution is emerging more slowly. Consumers already trust AI to summarize reviews, compare options, and recommend a best choice. But that trust drops sharply when AI moves towards autonomous action – prefilling baskets, completing checkout, or reordering without explicit approval.
This gap isn’t about technical capability. It reflects a behavioral threshold. Consumers want transparency, control, and the ability to intervene.
Trust will build as systems become more reversible, clearly authorized, and easy to audit.
For companies, this makes agentic commerce as much a customer experience challenge as a technology one. The brands that will earn trust fastest will be those that make interactions intuitive, collaborative, and transparent – while remembering past interactions and context.
New battlegrounds for growth
Taken together, these two shifts – internal agentic execution and external agentic decision-making – are reshaping how growth happens.
Inside the organization, value comes from redesigning workflows around agents. Outside, it depends on how well a company shows up in the AI-mediated environments where decisions are increasingly being made. Both need to be addressed in parallel.
Reshaping the enterprise for an agentic future
To unlock the internal value of agentic AI, organizations need to move beyond pilots and commit to transformation. Six steps stand out:
- Reimagine workflows as AI-first, domain by domain, starting from desired outcomes.
- Create a new human-agent frontier, with 75 percent of roles needing to evolve to include more technological, social, and cognitive skills.
- Shift to leaner, flatter organizational models built around autonomous “human + agent” teams.
- Drive change from the top, with clear leadership narratives that build confidence and adoption.
- Move every employee beyond AI fluency to daily integration of AI into their role.
- Rethink workforce planning to include both humans and agents as core resources.
These are the foundations of turning AI from a tool into an operating model.
Winning visibility in an AI-mediated market
At the same time, companies need to adapt to how customers discover and decide in an agent-driven market. That means focusing on a different set of priorities:
- Remaining visible in AI-mediated discovery environments where agents, not users, are doing the searching.
- Structuring product, pricing, and performance data so it can be easily interpreted and used by AI agents.
- Understanding how AI systems rank, compare, and recommend options – and shaping inputs accordingly.
- Building trust into agent-led interactions through transparency, control, and continuity.
- Designing experiences that work both for human users and the agents acting on their behalf.
This is not just a marketing challenge – it’s a commercial model shift.
Closing the gap between AI promise and commercial impact
The gen AI paradox does not resolve itself. The way forward is not more pilots, or forcing AI into processes that were never designed for it. It’s about rebuilding how work gets done around a human-agent model, and recognizing that the customer journey is being reshaped just as quickly.
Growth will increasingly depend on two capabilities: how effectively organizations deploy agents internally, and how well they compete in a world where the first customer interaction may not involve a human at all.
