Until recently, agentic AI in commerce has largely lived in the pilot phase. Businesses tested AI-powered agents for increasingly complex tasks, like generating product pages, drafting and localizing product descriptions at catalog scale, and managing promotions. Those pilots have proven their point. Now, AI adoption is no longer a differentiator. The question has shifted from “Does this work?” to “How do we do this at scale?“
According to the newly released State of Commerce report, more than a third of agentic AI users say their primary focus has shifted from pilots and experimentation to scaling AI across functions and teams. An inflection point is here.
We surveyed over 3,400 commerce leaders to learn how they’re shifting their mindsets and strategies to adapt to a new reality where generative AI is a core operating capability and no longer an experiment.
Pressure mounts to show AI results and return on investment
The most common AI agent implementations today are still customer-facing: autonomous customer service resolution, AI shopping concierges, and exception handling for returns and post-purchase interactions. These use cases make sense as a starting point; they offer faster feedback loops and more visible ROI than back-office deployments. But the footprint is widening. Organizations are increasingly finding success applying AI agents to more complex operations like supply chain management and merchandising optimization.
As the scale of AI increases, so does the pressure to show results. Pilots can run on goodwill and curiosity, but full-fledged deployments can’t. Once AI moves from a single team’s experiment to an investment spanning functions and budgets, leaders need to demonstrate the payoff. This means tying AI initiatives to metrics like conversion, retention, cost-to-serve, and revenue per interaction. Organizations that can clearly show ROI early are the ones best positioned to secure continued investment and expand into the next use case. Those that can’t risk stalling out, no matter how promising the underlying technology is. And this all hinges on data.
The catch: Scaling exposes what pilots hid
More than 6 in 10 organizations cite poor data integration (63%), lack of a defined AI strategy (63%), and poor data quality (62%) as major or moderate barriers to AI success. These aren’t new problems introduced by AI. They’re foundational issues that small-scale pilots were simply too limited to expose. When you’re running a single proof of concept, messy data might be a surmountable annoyance. But when you’re scaling AI across functions, teams, and channels, it becomes the ceiling on what you can achieve.
Clean, unified data is the real foundation for AI at scale
AI is only as good as the data feeding it. Right now, only 27% of organizations report having fully unified customer data across sales, service, marketing, and commerce teams. And that gap matters a lot. Disconnected data negatively affects the entire customer journey, from acquisition to conversion and loyalty.
On the other hand, organizations that have unified data report 40% better AI and automation outcomes and 40% stronger customer retention. That’s because you can’t personalize at scale without a single, accurate view of the customer. You can’t automate confidently without clean, connected inputs. Fixing data fragmentation isn’t a separate IT initiative that competes with AI investment for budget and attention — it’s the prerequisite that determines how much value that AI investment can actually deliver. Organizations serious about scaling AI need to treat data unification as step one, not a project to revisit later.
How are businesses building for what’s next with AI?
If the barrier to scaling AI is a foundation problem, the good news is that commerce leaders are actively working to improve it. Data unification is a clear focus area: Improving data quality, accessibility, and management is a top priority for organizations this year.
But data unification isn’t a single project. It’s a set of tactical, often unglamorous tasks. Commerce leaders point to a few in particular:
- Consolidating systems of record. Rather than layering AI on top of five different customer databases, teams are prioritizing a single source of truth for customer, order, and inventory data before expanding AI use cases further.
- Fixing pricing and inventory sync issues. 98% of multi-channel sellers report significant omnichannel breakdowns — inconsistent pricing (37%) and inventory sync problems (37%) chief among them. Cleaning up these specific data flows is one of the most immediate, high-leverage fixes teams are making, since they directly undermine both the customer experience and any AI built on top of that data.
- Auditing integration points between core systems. With poor data integration cited as the top barrier to AI success (63%), teams are going system-by-system to identify where data breaks or duplicates as it moves between platforms — commerce, CRM, order management — rather than assuming the AI layer will smooth it over.
- Defining ownership of the data foundation. Because data quality (62%) and lack of AI strategy (63%) are both top-cited barriers, teams are increasingly assigning explicit ownership over data readiness, rather than treating it as a shared responsibility that no one is accountable for.
- Building the metrics before scaling further. Even as unification work continues, leaders are starting to define the specific KPIs they’ll use to justify the next round of AI investment, rather than scaling first and measuring later.
Organizations treating unification and integration cleanup as the current priority are the ones best positioned to make the leap from promising pilot to dependable, revenue-driving scale.
Impactful, scalable AI with Agentforce Commerce
Agentforce Commerce is an AI-first platform that makes it easy to close the data gap. Rather than asking businesses to solve their data problems and develop their AI strategies separately, Agentforce Commerce is designed to reduce the infrastructure lift that’s currently the single biggest barrier standing between AI pilots and real, sustained scale. Instead of adding another disconnected tool to an already sprawling stack, Agentforce Commerce gives commerce teams a way to deploy trusted, enterprise-grade AI agents on a foundation that’s already built for it, from customer service resolution to merchandising and order management, all working from the same data.
For organizations still trying to figure out how to move from experimentation to enterprise-wide scale, that foundation is the difference between AI that shows flashes of promise and AI that reliably delivers.
Read the full State of Commerce Report to explore more data >>
