An AI Companion for Adults With ADHD Is a Conversation Tool, Not Treatment

Separate companionship from ADHD care

NIMH’s ADHD overview describes ADHD as a developmental disorder and routes diagnosis and treatment to qualified health professionals. It does not recommend conversational agents. An AI companion may be used for low-stakes conversation, but this page makes no claim that the conversation changes symptoms or underlying biology. Medication, diagnosis, treatment and urgent questions remain outside the chat. The audit below records only whether a reader-defined physical action followed a prompt.

Failure mode appears when the user delegates an important task or health decision to the conversation. Decision action: label the intended job before opening the chat and stop if the prompt becomes a request for diagnosis, medication advice or a substitute for professional care.

Choose one low-stakes friction point

Begin with one observable, low-stakes barrier rather than vague overwhelm. Examples include opening a personal spreadsheet, drafting a non-urgent email or putting one item away. The reader decides what “low stakes” means. Do not use the companion for an action whose delay could affect health, safety, legal rights, employment or significant money.

Mechanism: naming the exact physical transition (mouse click, foot movement, lid closure) converts an abstract executive-function complaint into a testable micro-behavior. Evidence boundary: no published study demonstrates that conversational rehearsal alone produces lasting change in ADHD-related initiation; therefore the audit records only what the user actually does after the prompt, never infers clinical improvement.

Complete the ADHD Companion Friction Audit

The audit is a blank reader-run documentation tool. Copy the table below into a private note-taking app and fill one row per test. It contains no symptom score, diagnosis, treatment result, or productivity percentage.

Run the audit only when it answers a specific low-stakes question. After a row, compare the expected action with what actually happened before deciding whether another prompt is useful. The healthy AI companion boundaries page supplies additional questions for separating conversation from reliance.

Test prompts without delegating important decisions

Prompts function as verbal scaffolding, not decision engines. A safe test prompt names the exact micro-step, caps the model’s reply length, and ends with an instruction to stop. Example: “Describe only the first keystroke needed to open my invoice template. Do not add motivation, metaphors, or follow-up questions.” After the model answers, the user must execute the described step without further chat.

Failure mode: open-ended prompts such as “Help me focus” do not define an observable result. If the conversation expands while the intended action remains unchanged, close the tab and reconsider the prompt. The companion cannot own the real-world outcome.

Control notification, novelty and spending loops

Notifications or an always-open tab may introduce another distraction. Test one setting at a time and record what changed; do not assume notification removal improves ADHD symptoms. A prompt may request a short answer without follow-up questions, but the reader chooses the response boundary and stop cue.

Treat a Pass as an entertainment purchase, never a health investment. AISoul’s fixed non-renewing prices and free allowance are commercial constraints, not clinical dosing. Before buying, record the live price, access duration, purpose and a reader-defined budget. The number of completed audit rows does not prove value or health benefit.

The AISoul Privacy Policy lists stored data categories, external AI providers and the absence of an end-to-end-encryption claim. Exclude identifiers, medical records, credentials and financial details that the task does not require. This is data minimisation, not a promise that the remaining chat is anonymous.

Keep treatment and urgent support human

An AI companion for ADHD adults must never serve as a crisis line, therapist, or medication manager. The NIMH NIMH mental-health and help guidance page distinguishes routine self-care from professional and urgent support and does not designate chatbots as substitutes. If a conversation reveals thoughts of harm, severe distress, or worsening symptoms, the only correct next action is to contact a licensed clinician, crisis service, or trusted human contact.

The AI companion versus therapist boundary separates casual conversation from professional support. Do not wait for a generated disclaimer to decide: route medication, diagnosis, treatment and urgent-safety questions to a suitable human source from the outset.

Questions about AI companions for adults with ADHD

Can an AI companion treat adult ADHD?

No. Diagnosis and treatment remain the domain of qualified health professionals. An AI companion supplies conversational rehearsal of small tasks but supplies no therapeutic mechanism, cannot prescribe or adjust medication, and carries no medical-benefit claim. NIMH guidance routes all clinical questions to licensed providers; any perception of symptom relief must be verified by a clinician and not attributed to the chat tool.

Can it help me start a small task?

It can supply a bounded verbal prompt that names the first physical micro-step. The user must still execute that step in the real world. The audit table records whether the expected action occurred; repeated failure after prompting indicates the need to simplify the task further or consult a professional for strategy adjustment. The companion never owns the outcome.

Are reminders always useful for ADHD?

No universal rule exists. Some adults find external reminders helpful; others experience notification fatigue that increases avoidance. Test one notification setting at a time inside the friction audit, observe the actual next action, and adjust. Reminders from a non-human source also lack the social accountability that human check-ins can provide; therefore they remain an experiment, not a standard.

What information should I keep out of the chat?

Keep all personally identifiable, medical, financial, legal, or otherwise sensitive data out of every conversation. The service stores chat logs, uses external AI providers, and makes no anonymity or absolute-security guarantee. A practical rule is: if you would hesitate to post the information on a public social-media account, do not enter it here. Review the current privacy policy before each new thread.

When should I speak with an ADHD professional?

Speak with a professional whenever the friction audit shows consistent failure to execute recorded physical actions, when distress escalates, when medication or therapy questions arise, or when daily functioning is meaningfully impaired. The companion can surface the question but cannot answer it. Use the clinician-owned column in the audit to log the exact concern, then book an appointment rather than continuing the chat.

NIMH scope, product limits and conflict disclosure

Evidence routes for reproducing this page's checks: AI companion versus therapist boundary; AISoul Privacy Policy.

NIMH materials describe ADHD as a developmental disorder and distinguish self-care from professional treatment; they do not test, endorse, or supply usage protocols for any AI companion. Product limits at AISoul—daily message caps, fixed non-renewing passes, and finite gallery access—are commercial decisions published on the vendor page and do not constitute health recommendations. Free access resets on a Beijing calendar day; these quotas are not clinical thresholds.

This article is published by AISoul, which develops and sells the adults-only browser companion described and therefore has a commercial interest. Claims are bounded by the two cited NIMH pages; neither tests or endorses an AI companion. The Friction Audit is personal documentation, not a diagnostic or therapeutic tool. Current policy must be checked live, while medical decisions remain with qualified health professionals.