SimplePractice and TherapyNotes both now draft treatment plans with AI. Medicare's own coverage rules still name six specific elements neither tool is built to guarantee.

Two of the largest EHR platforms serving independent mental health practices shipped competing AI treatment plan tools within weeks of each other this year. SimplePractice launched Care Aide, an add-on that includes a Treatment Planner feature, on 17 June 2026 (SimplePractice, via BusinessWire, "SimplePractice Launches AI-powered Care Aide," 17 June 2026). TherapyNotes already offers TherapyFuel, a $40-per-clinician-per-month add-on that generates draft treatment plans with goals, objectives, and interventions tied to the diagnoses already in the chart (TherapyNotes support documentation and TherapyNotes blog, "Supercharge your Documentation with TherapyFuel," checked August 2026). Both are marketed as time-saving documentation tools. Neither vendor mentions, in its own marketing copy, that the treatment plan is the one clinical document Medicare's coverage rules check against a named list of required content, not just whether goals and interventions are present.

This matters for an independent psychology practice for a reason that has nothing to do with clinical judgment and everything to do with getting paid. A progress note that reads a little generic rarely gets challenged. A treatment plan missing a required element is a documentation finding a payer can act on, during a routine audit or a continued-stay review, months after the AI draft was accepted and filed.

What Medicare's coverage rule actually checks

Medicare pays for outpatient psychotherapy under coverage rules set by each region's Medicare Administrative Contractor, published as Local Coverage Determinations. Several of these LCDs for psychiatry and psychology services describe a consistent set of required treatment plan content, summarized by AAPC's coding compliance guidance and corroborated across the LCDs themselves (Centers for Medicare and Medicaid Services, Local Coverage Determinations for Psychiatry and Psychology Services, multiple contractor jurisdictions; AAPC, "Meet Documentation Requirements for Psychotherapy Services," checked August 2026). Six elements recur: the target symptoms being treated, the goals of therapy and the method used to monitor progress toward them, a stated reason the chosen therapy is the appropriate treatment rather than or alongside another approach, an estimated duration of treatment expressed as a number of sessions, documentation that the patient has the capacity to participate in and benefit from psychotherapy, and ongoing evidence of medical necessity as treatment continues past the initial sessions.

One caution worth flagging directly: LCDs are set by the local Medicare Administrative Contractor, not by a single national rule, so exact wording and emphasis vary somewhat by jurisdiction. CMS.gov's own LCD pages require a JavaScript-rendered session this research could not fully retrieve, so the six-element summary above is drawn from a search-engine synthesis of several published LCDs (including L39853 and L33632) alongside AAPC's coding compliance guidance, not a direct read of the primary LCD text itself. The pattern is consistent across every source checked. Treat the six elements as the reliable core, and confirm your own MAC's specific LCD wording before relying on this for an active audit response.

So what this means for you: a treatment plan built only from a diagnosis, a handful of goals, and a list of interventions, which is exactly the shape most AI treatment plan generators produce by default, is missing at least three of the six elements a Medicare reviewer is trained to look for. None of that is a reason to avoid these tools. It is a reason to know what to add before filing.

What the two new AI tools actually generate

SimplePractice's Care Aide add-on costs $59 a month for the account owner and $49 a month for each additional clinician in a group practice (SimplePractice Support, "Introducing Care Aide," updated 22 July 2026). Its Treatment Planner feature is described by the company as helping "create treatment plans faster, reducing manual reviewing of clinical documentation, progress notes, and outcome measures, while keeping the clinician in control of care decisions" (SimplePractice, via BusinessWire, 17 June 2026). SimplePractice's own support documentation is direct about the tool's limits: "the content may be incomplete, inaccurate, or missing context. It does not replace your clinical judgment or serve as final documentation," and clinicians are instructed to "always review and edit all AI-generated content before using it in practice" (SimplePractice Support, "Introducing Care Aide," updated 22 July 2026).

TherapyFuel works from existing chart data to draft goals, objectives, and interventions for each treatment focus, matched to the diagnoses already documented (TherapyNotes support documentation and TherapyNotes blog, checked August 2026). Independent reviews describe the drafts as structurally sound but note that clinicians still need to adjust language for diagnostic nuance, risk documentation, and medical necessity, and that the system does not track how well a plan aligns with actual session content over time (Mentalyc, "TherapyNotes AI (TherapyFuel) Review for Therapists," checked August 2026, an independent vendor-review site, not a primary source; flagged accordingly). TherapyNotes' own guidance states that generated plans are drafts for the clinician to edit and approve, and that notes are never auto-finalized by the AI.

So what this means for you: both tools are explicit, in their own documentation, that they generate a starting draft rather than a finished, payer-ready plan. That framing matches what the tools are actually built to do. What it does not do is tell a solo or small-group practitioner which specific fields to add before that draft becomes the version filed in the chart.

The professional guidance says review it. It does not say what to check for

The American Psychological Association released Ethical Guidance for AI in the Professional Practice of Health Service Psychology in June 2025, its first formal position on the topic (American Psychological Association, cited via Blueprint, "ACA and APA Ethical Guidance for the Use of AI for Therapists," published 13 May 2026; direct fetch of apa.org was blocked by the site's bot protection this session, so this is drawn from a secondary summary corroborated independently against a separate search of APA's guidance). The guidance places weight on validation, accuracy, and human oversight before any AI tool is adopted into practice, and is explicit that the clinician, not the software, remains responsible for every clinical decision and for reviewing AI output before it is acted on or filed.

That guidance is a professional body's ethical position, not a licensing requirement with the force of law, and it does not name Medicare's six-element checklist, TherapyFuel, or Care Aide specifically. It tells a practitioner to review AI output carefully. It does not tell a practitioner what a Medicare reviewer is actually going to check for. The gap between "review it" and "know what a payer's coverage rule requires it to contain" is exactly where an AI-drafted treatment plan that reads well can still fail an audit.

So what this means for you: neither the APA's ethical guidance nor either vendor's own product documentation is going to hand a practice the specific list a Medicare reviewer works from. That list exists, it is stable, and it takes a few minutes to check a draft against once you know what it is.

A checklist worth keeping next to the AI draft

Before filing any AI-drafted treatment plan for a Medicare patient, or for a commercial payer using comparable medical-necessity standards, confirm the draft states the target symptoms in the patient's own presenting language, not just a diagnosis code. Confirm it names the specific goals of therapy and how progress toward them will be measured, not only a list of interventions. Confirm it gives a reason the chosen therapy is the right approach for this patient, particularly if more than one modality is plausible. Confirm it states an estimated duration of treatment in a number of sessions, not an open-ended plan. Confirm it documents the patient's capacity to participate in and benefit from psychotherapy, a specific clinical judgment that has to come from the clinician, not the software. And confirm the plan gives a reviewer, reading it cold, a clear basis for ongoing medical necessity, not just a snapshot of where treatment started.

Most of that is a five-minute addition to a plan an AI tool has already drafted correctly on the parts it was built to draft. The six-element check does not replace clinical judgment. It replaces the assumption that a well-written AI treatment plan and a Medicare-compliant one are automatically the same document.

The call

If your practice is already using or evaluating TherapyFuel, Care Aide's Treatment Planner, or any comparable AI drafting tool, keep the tool. The time savings on goals, objectives, and interventions are real, and both vendors are transparent that a clinician has to review and approve every output before it becomes part of the record. Add one workflow step: before a Medicare or Medicare-adjacent treatment plan is filed, run it against the six elements above, specifically the duration estimate and the capacity-to-benefit statement, since those are the two most likely to be missing from a template built around goals and interventions alone.

If clinical governance and documentation risk are the reason your practice has been cautious about adopting any AI tool at all, that is exactly the kind of gap the AI Opportunity and Growth Assessment is built to surface against your own payer mix, not a generic vendor demo. Start with a free 20-minute discovery call.

See also: our look at five AI use cases in psychology practice that skip clinical notes entirely, how to weigh a general-purpose AI tool against a purpose-built clinical scribe, and what informed consent requires before any AI tool touches a client record.

Related: Psychology groups: 5 AI use cases that skip clinical notes · $20 vs $150: ChatGPT or an AI scribe · Psychology AI already runs across 550 locations