Introduction
As organizations expand the use of artificial intelligence (AI) across business operations, many are discovering that deploying AI agents requires a different management approach than traditional software systems. Unlike rule-based automation, AI agents are designed to analyze objectives, evaluate available information, and recommend methods for completing complex tasks.
According to Richard Kent of TAINA Technology, businesses may limit the potential of AI agents by treating them like conventional software that follows predefined instructions. Instead, organizations may achieve better outcomes by defining objectives, providing context, and allowing AI systems to propose workflows while maintaining human oversight and accountability.
What Are AI Agents?
AI agents are software systems capable of performing tasks with varying degrees of autonomy. Rather than simply executing predefined commands, they can interpret goals, identify missing information, evaluate alternatives, and recommend actions based on available data.
In professional services, AI agents are increasingly being explored for activities such as:
- Research and information gathering
- Large-scale data analysis
- Report generation
- Workflow automation
- Document preparation
- Process optimization
Unlike traditional automation tools, AI agents can adapt their approach when circumstances change, making governance and oversight essential components of enterprise deployment.
How AI Agents Differ From Traditional Software
Conventional enterprise software typically operates according to fixed rules and structured workflows created by humans. Every process step is defined in advance, and the software performs tasks exactly as instructed.
AI agents operate differently.
Instead of following rigid workflows, they can:
- Interpret business objectives.
- Identify information gaps.
- Evaluate dependencies.
- Recommend alternative approaches.
- Optimize workflows based on available context.
This capability allows AI agents to contribute beyond task execution, but it also requires organizations to rethink how they interact with intelligent systems.
Why Asking Questions Can Produce Better Results
Richard Kent argues that organizations often begin AI projects by attempting to map every stage of a workflow before introducing an AI agent.
During the development of an agent-based software development workflow, Kent initially created a detailed process covering documentation, approvals, checkpoints, and information flows. Although the AI agent successfully followed the prescribed workflow, the resulting process reflected human assumptions rather than opportunities for optimization.
Kent then changed the approach by giving the AI agent a clear objective and asking two key questions:
- What information is required to complete the task effectively?
- How should the workflow be structured to achieve the objective?
The AI agent responded by identifying missing business requirements—including stakeholder objectives, testing expectations, governance controls, deployment constraints, and feedback mechanisms—and proposed a more iterative workflow supported by continuous validation and automated quality checks.
The experience illustrates how AI agents may generate additional value when organizations encourage analysis rather than prescribing every procedural step.
Human Oversight Remains Essential
While AI agents can recommend workflows and automate complex tasks, responsibility for business decisions remains with human professionals.
Organizations continue to retain accountability for:
Governance
Ensuring AI operates within organizational policies and regulatory requirements.
Compliance
Reviewing outputs for legal, financial, and industry-specific obligations.
Risk Management
Assessing recommendations before implementation, particularly where business or regulatory risks are involved.
Final Decision-Making
Approving actions that could affect customers, financial reporting, or regulatory compliance.
Many organizations are adopting “human-on-the-loop” operating models, where AI performs analysis while people supervise outcomes and intervene when necessary.
Implications for Professional Services Firms
Professional services firms—including tax, accounting, legal, consulting, and audit organizations—are increasingly evaluating AI agents for knowledge-intensive work.
Potential applications include:
- Tax research
- Due diligence reviews
- Risk identification
- Workpaper preparation
- Regulatory documentation
- Client reporting
- Data validation
Rather than replacing professional expertise, AI agents may shift employee responsibilities toward defining objectives, evaluating outputs, and exercising professional judgment.
This change places greater emphasis on critical thinking, governance, and oversight alongside technical AI capabilities.
Challenges and Limitations
Despite their flexibility, AI agents present several implementation challenges.
Organizations must consider:
- Data quality and completeness.
- Transparency of AI-generated recommendations.
- Regulatory and compliance requirements.
- Information security and privacy.
- Integration with existing enterprise systems.
- Monitoring AI performance over time.
In regulated industries, organizations may also need to demonstrate how AI-assisted decisions were made and maintain auditable records for internal governance and external regulatory reviews.
Future Outlook
As enterprise AI adoption continues to expand, organizations are expected to place increasing emphasis on collaboration between humans and intelligent systems rather than full automation.
Industry observers suggest that successful AI implementation will depend not only on technological capability but also on governance frameworks, workforce skills, and operational processes that clearly define human responsibility.
Businesses that combine AI-generated analysis with structured oversight may be better positioned to integrate autonomous systems into complex professional workflows while maintaining accountability.
Conclusion
AI agents represent a shift from traditional software by contributing analysis, recommendations, and workflow optimization in addition to task execution. Richard Kent’s analysis suggests that organizations may achieve greater value by defining business objectives and engaging AI agents through structured collaboration rather than prescribing every operational step. As AI capabilities continue to evolve, effective governance, transparency, and human oversight are likely to remain central to enterprise adoption.
