Enterprises across the wholesale, retail, and construction sectors are increasingly integrating artificial intelligence into their daily operations while maintaining a strict "look-up only" policy for AI agents. According to the PYMNTS Intelligence report, “Wholesale Writes the AI Playbook,” no company in these sub-industries currently permits autonomous AI action, opting instead to restrict agents to information retrieval to mitigate risks regarding data exposure and unauthorized decision-making.
Why Companies Restrict AI Agents to Look-Up Access
While AI adoption is rising, businesses are drawing a clear line at execution. Data from a May survey of 60 senior technology executives at U.S. enterprises with at least $1 billion in annual revenue shows that 100% of wholesale firms, 90% of retailers, and 85% of construction firms limit AI agents to retrieval tasks.
This restriction is a deliberate governance strategy rather than a technical limitation. Executives remain concerned about the lack of traceable identities for AI agents and the potential for systems to operate outside of established corporate boundaries. By limiting agents to look-up access, firms can leverage AI for research, document generation, and threat monitoring without risking the immediate, automated execution of business transactions.
How AI Adoption Varies by Sector
Deployment levels differ significantly across the goods economy. Wholesale firms currently lead in maturity, utilizing AI across 35 of the 75 business tasks tracked in the survey.
- Wholesale: Firms show the highest deployment, with majority usage in six of eight business functions. Key applications include contract generation and security monitoring, both at 75% adoption.
- Retail and Construction: These sectors utilize AI across approximately 31 tasks each, with a primary focus on marketing and sales-related activities.
Despite this broad integration, the "look-up" ceiling remains consistent. Only 5% of retail firms allow any form of execution, and this is strictly contingent upon human approval. In contrast, 15% of construction firms grant AI agents no access to internal systems at all.
What is the Five-Year Outlook for Autonomous AI?
The industry consensus for the next five years is moving toward semi-autonomous operation rather than full automation. Research indicates that 85% of wholesale firms and a majority of retail and construction companies identify semi-autonomous AI—defined as AI systems that require human oversight—as their primary long-term goal.
No firms surveyed identified fully autonomous AI as their vision for the near future. This shift reflects a broader business process challenge: trust. Previous PYMNTS Intelligence findings revealed that 98% of product leaders do not feel prepared to grant core-system access to fully autonomous agents.
However, the potential benefits of moving beyond simple look-up access are significant. Firms that successfully implement AI agents capable of autonomous action have reported automating up to 95% of accounts receivable tasks, compared to just 38% for firms that rely on manual or limited-access AI processes.
Key Takeaways
- Uniform Restriction: Every wholesale firm surveyed limits AI agents to look-up access, the most consistent posture across all sub-industries.
- Governance Over Capability: The refusal to grant execution rights is rooted in risk management, specifically concerns regarding sensitive data exposure and lack of accountability.
- The Trust Gap: Despite high levels of AI deployment, business leaders cite trust as a larger barrier to full adoption than the underlying technology.
- Future Goals: Most firms are targeting semi-autonomous workflows with human intervention as their five-year strategic objective, rather than complete automation.