Organizations are experiencing a surge in employee-built AI agents as workers leverage these tools to streamline workflows and solve specific business challenges.
At West Monroe, CAIO Bret Greenstein holds weekly meetings where employees showcase their custom AI solutions—more than 550 agents have been created by 200+ staff since May alone. Similar trends are emerging at Blackline, Microsoft (with over 500,000 internal agents), and across Fortune 500 companies.
These homegrown agents address critical pain points that enterprise software often misses—from automating invoice processing to providing sales teams with targeted customer insights.
However, this proliferation presents new governance challenges. Michael Murphy of Adaptovate notes a shift from initial enthusiasm to concerns about agent value and cost control. CIOs now face pressure from multiple directions:
- Employees seeking broader AI access
- CFOs concerned about rising token expenses
- CEOs demanding ROI from AI investments
To address this, experts recommend:
- Implement role-based access controls: Allow usage while ensuring corporate standards are met.
- Treat agents like employees: Grant specific credentials and monitor activity through audit logs.
- Utilize sandboxing: Contain agents within secure environments with limited data access.
- Establish an approval process: Require requests for new agent identities through a centralized system
By proactively establishing these frameworks, organizations can harness the innovation potential of employee-built AI while mitigating risks and ensuring alignment with business goals.