Operational AI, Part 1: The Biggest AI Opportunity Is in Your Back Office

If you’ve followed the conversation around artificial intelligence (AI) in behavioral health, you’ve probably heard about AI therapists, chatbots, and clinical decision support tools. Those are the examples that make headlines, drive legislation, and spark debate, but they represent only a small portion of the AI solutions available in behavioral health.

For most behavioral health organizations, one of the biggest opportunities is around reducing administrative burden, strengthening financial performance, improving compliance, streamlining patient communication, and giving staff back time so they can focus on what truly matters: taking care of their clients and themselves.

AI is not just one thing; it’s both/and. There are AI-powered tools that your organization may deem high risk to implement, and there are operational AI solutions that can drive tremendous improvements in your organization that are worth implementing.

What Is Operational AI?

Welcome to Part 1 of our four-part series on operational AI:

1. The Biggest AI Opportunity Is in Your Back Office: How Operational AI Solutions Can Drive Performance
2. The DSM of Operational AI: A Practical Way to Organize the Operational AI Landscape (coming soon!)
3. Beyond Asking AI About AI: How Behavioral Health Leaders Find AI Solutions (coming soon!)
4. What the Demo Won’t Tell You: How to Evaluate Operational AI Vendors (coming soon!)

Just in case you’re wondering, what exactly is “operational AI”? It’s generally viewed as AI that focuses on business/back-end workflows that are defined by certain rules.

Note: This article will not name specific vendors but rather areas in which you can implement nonclinical, operational AI solutions today.

Key Opportunities for Operational AI in Behavioral Health

Revenue Cycle Management

AI tools for revenue cycle management can review documentation for coding accuracy, identify claims that are likely to be denied before submission, assist with revenue cycle workflows, monitor payer requirements, correct coding to reduce the number of denials, and help staff prioritize any remaining denials.

We know that margins are thin, so reducing preventable denials and accelerating reimbursement is an excellent initial way to leverage AI.

Example: For many National Council member organizations, the financial side of AI is delivering the quickest return on investment, capturing anywhere from 15%-40% of additional revenue due to incorrect coding or denials.

Compliance and Quality Improvement

Instead of manually reviewing a small sample of charts, AI-powered compliance solutions can examine documentation across the entire organization to flag missing elements, identify compliance concerns, and surface trends that deserve further review. The tools can even aggregate compliance and reporting criteria across the state, health plans, and other regulatory bodies and analyze which of those criteria are or are not being met.

All this automated work doesn’t replace your compliance staff or put compliance on autopilot. Instead, it helps your staff focus their expertise on finding solutions for those issues and working with the rest of the team to implement them.

Automating Repetitive Administrative Work

Many administrative tasks follow predictable workflows that require attention but not necessarily complex judgment. AI tools can help verify insurance eligibility, prepare prior authorization requests, manage appointment reminders, organize waitlists, route referrals, draft routine correspondence, and support countless other administrative processes.

Individually, these tasks may seem small. Collectively, they account for thousands of staff hours every year that can be reattributed to work that better suits their capabilities and licensure.

Accelerate Administrative Writing

One of AI’s simplest and most accessible uses is helping staff create first drafts. Policies and procedures, board reports, grant narratives, job descriptions, meeting summaries, communications, training materials, and internal documentation all require time to write.

Instead, staff can:

• Populate AI with the criteria and requirements of the document being written
  • Note: AI does not always pull the most up-to-date data; ensure you are prompting the AI to conduct present-day web searches, so that it doesn’t just pull information from a stored historical database.
• Prompt AI to produce an accurate and straightforward draft while also citing areas of weakness.
• Focus their time on reviewing and customizing the content to their specific work environment, rather than starting from a blank page.
  • Privacy note: Be careful what organizational level or personal information is being shared with the AI. (Cue the annual PHI, privacy, and security training!)

Example: We’ve heard from several members that staff are using AI to write policies, grants, and other important documentation that can be relatively straightforward. As a result, they have saved anywhere from a handful up to 20 hours a week.

Data-driven Decision-making

Behavioral health organizations collect enormous amounts of data, but transforming that information into meaningful insight often requires significant analytical work. AI can help summarize performance trends, build dashboards, identify emerging patterns, and answer operational questions such as:

• Which programs have the highest no-show rates?
• Where are documentation bottlenecks occurring?
• Which services are experiencing growing demand?
• Which grant metrics are trending off target?

Instead of simply reporting what happened, AI can help leaders understand where to focus their attention next.

Getting Started With Operational AI

You do not need to develop an enterprise-wide AI strategy before testing a particular operational AI application. A good place to start is to ask yourself, Which operational problem is consuming the most time, creating the most sustainability risk, or limiting our ability to serve people? My guess is that you are likely to say:

• High denial rates
• Missed revenue capture
• Slow prior authorization
• Long intake delays
• Clinician documentation burden
• Credentialing bottlenecks
• Repeated reporting requirements
• Inconsistent chart review
• Preparing grant submissions
• Compiling data for states, plans, and funders

Those are often the areas where AI can create value right away. Once that is clear, leaders can evaluate whether an AI-enabled solution is appropriate, whether the necessary data and workflows are available, and what level of human oversight is required.

Conclusion

As conversations about AI continue to evolve, it’s worth remembering that the future of AI in behavioral health isn’t just about what happens in the therapy room. It is also about everything that happens behind the scenes to make your organizations and teams more successful and sustainable.

Author

Anjlee Joshi
Principal Consultant, Impact Investing Initiative
National Council for Mental Wellbeing