AI Becomes Mental Health's Chief of Staff - With Humans at the Helm

Meet the New 'Chief of Staff' in Mental Health: AI

Artificial intelligence has moved from side conversations to the center of executive workflows. Generative tools like ChatGPT, Gemini, Claude, and enterprise copilots now sit beside leaders as thought partners, speeding up analysis, planning, and communication.

A 2025 report found that 72% of VP and C-suite leaders use AI daily, often four times more than frontline staff. In behavioral health, that translates into lighter admin load, faster decision prep, and clearer strategic focus.

"AI is like a chief of staff of sorts, where I still have to be the editor and the final decision maker, but it has definitely streamlined things," said Howard Barker, executive director of business development at Your Behavioral Health. "It frequently gives better outputs than a lower-level executive assistant could for bouncing ideas - but you can't let it run wild."

How behavioral health leaders are using AI right now

• Remove friction and paperwork: first-draft emails, summaries, briefs, and proposals.
• Prep for meetings: agenda shaping, stakeholder maps, risk checklists, challenge questions.
• Structure programs and processes: draft SOPs, compare options, surface trade-offs.
• Real-time pulse checks: regulations, payer dynamics, funding shifts, acuity trends.
• Data-driven resource moves: demand analysis, occupancy insights, regional signals.
• Clinical ops support: EMR note assistance, documentation QA, med interaction flags, billing support.
• Staffing workflow pilots: census prediction, scheduling suggestions, coverage modeling.

Prompting is now an executive skill

Dr. Kathryn Boger, co-founder and chief clinical officer at InStride Health, treats prompting like a craft. Before any task, she asks where AI can add leverage. She often assigns AI a role - "consultant" or "expert assessment consultant" - to pressure-test strategies and standardize decision inputs while keeping human judgment in the loop.

Her team piloted an AI assistant to weigh in on clinical fit decisions and saw strong results, with plans to scale. "AI can be a sycophant," Boger said, "so I ask it to challenge my assumptions and surface risks or blind spots."

• Set a role: "Act as a healthcare strategy consultant for behavioral health…"
• Define the context: goals, constraints, data available, stakeholders.
• Force friction: "Challenge my plan. List blind spots, risks, failure points."
• Validate: cross-check outputs, cite sources, and keep final decisions human.

Want to operationalize this with your team? See Prompt Engineering for practical frameworks and team drills.

The tool stack executives actually use

• ChatGPT, Gemini, Claude: strategy draft work, research synthesis, tone calibration.
• Microsoft Copilot: email triage, daily priorities, language tuning for leadership comms.
• OpenEvidence: academic-first research for clinical and policy questions.
• Google NotebookLM: summarize datasets, auto-generate tables, produce quick draft slides and demo assets.

Example: A partnerships lead used NotebookLM to turn real client data into a five-minute, on-brand video and data view - created on the spot - while keeping tight control of inputs.

Where AI is moving the needle

Rob Marsh, CEO of Bradford Health Services, uses AI for demand analysis across SUD care. By pulling claims, occupancy, and regional indicators together, his team can reallocate beds and staff far faster than traditional reporting cycles.

Ascension Recovery Services is cutting documentation burden inside the EMR and using AI to spot relapse risk patterns, medication interactions, missing notes, and billing gaps. Predictive outcomes and early detection of return to use are on the near-term roadmap. Staffing suggestions tied to census shifts are in exploration.

At Eleos Health, leadership time now includes vetting AI use cases with a high bar for clinical rigor, security, and ethics. Internally, teams are encouraged to ship responsible prototypes and share what works, building compounding know-how across the org.

Non-negotiables: keep clinical judgment human

Leaders are clear: AI can guide, draft, and analyze - but clinicians and executives own the call. "There's always going to be a need for a steady hand at the wheel," Marsh said. The upside is better inputs and faster iteration, not abdication.

• Human-in-the-loop on all clinical impact areas.
• Privacy and PHI safeguards by default; limit data exposure and access.
• Bias checks and red-teaming for high-stakes workflows.
• Audit trails: what was generated, accepted, or edited - and by whom.
• Ongoing training on safe use, limitations, and decision boundaries.

A simple operating system for AI-augmented leadership

• Morning pulse: ask AI for 10 key shifts in regulation, payers, and local market signals; convert to a 3-bullet brief.
• Meeting prep: generate a one-page memo with goals, counterarguments, and go/no-go criteria.
• Decision support: have AI list second- and third-order effects, then pre-mortem the plan.
• Weekly: audit your top workflows and move one repetitive task to AI-assisted.
• Metrics: track cycle time saved, quality of decisions (leading indicators), and error rates.
• Governance: define what AI can do, what it must not do, and approval paths.
• Team enablement: build a prompt library and run short drills on real scenarios.
• Tool evaluation: score by data control, accuracy on your tasks, speed-to-value, and integration ease.

For leadership playbooks that map directly to executive work, explore AI for Executives & Strategy.

Favorite prompts leaders rely on

• "Act as a healthcare executive with behavioral health expertise. What risks, blind spots, and ethical concerns am I missing in this plan?"
• "Summarize these research articles and policy updates into a 6-bullet executive brief with citations."
• "Pressure-test this decision. List second- and third-order effects and where this could fail."
• "Draft a demand analysis plan using claims, occupancy, and regional data. Specify data sources, cadence, and thresholds for action."
• "Turn these meeting notes into a one-page action plan with owners, deadlines, and dependencies."
• "Rewrite this message for frontline staff: clear, concise, supportive tone; 150 words."

The leadership edge going forward

Executives will need two muscles: the ability to ask sharper questions of AI, and the judgment to evaluate tools on fit, data control, and impact. As Barker put it, new tools create new roles - if people evolve with them.

Think like a project manager, even if it's not your title. Keep the human judgment high and let AI do the grunt work.

Companies featured

• Ascension Recovery Services
• Aware Recovery Care
• Bend Health (a Lyra Health company)
• Bradford Health Services
• Eleos Health
• InStride Health
• Your Behavioral Health