Commitments

Clinicians do not read companion conversations.

No clinician or practice staff member can read a Client’s companion conversations or private companion memory, except the crisis-flag window described in the privacy policy.

The journal and private notes are the Client’s alone.

No staff, operator or AI reads them. If the Client dictates, Amazon’s speech service turns their voice into text. Optional passcode seal on the device, with no recovery. Backups keep earlier and deleted words up to 35 days.

Client Data is not used to train any model.

Neither Mojo AI Services LLC nor any service provider uses Client Data to train or fine-tune an artificial intelligence or machine learning model. This is a term of the agreement with the model provider and of the business associate agreement with each practice.

Client Data is not sold and is not used for advertising.

Operational measurements use counts, timings and error codes rather than conversation content. These measurements support service reliability and usage accounting.

Clinical storage and AI processing use United States regions.

AWS hosts the application records and AI processing in United States regions. Cloudflare also processes application traffic in transit, as described in the privacy policy.

The practice holds the record.

The practice controls its clinical record. Where HIPAA applies, the business associate agreement governs Mojo AI Services LLC’s processing of protected health information. A billing lapse does not remove records access or export. Data return, retention and destruction follow the practice agreement, legal holds and applicable law.

Data collection is limited to what the Services require.

The Services do not collect device contacts or precise device location. Optional dictation streams audio to Amazon Transcribe, an AWS service, to return text. Mojo AI Services LLC does not save it as an audio recording, and its AWS account is opted out of AWS AI services using that audio to develop or improve their services. Retention follows the practice policy and business associate agreement.

Product improvement does not grant permission to read Client conversations.

We improve the instructions, safeguards, evaluation cases and workflows around the AI models. Any separate review of Client content for product quality or research requires the applicable authorization and agreements. Accepting the privacy policy does not enroll a Client in such a program.

The Companion discloses that it is an automated system.

The Companion is identified as AI in the conversation interface and consent process. It is not a person or a licensed professional.

The Companion does not make diagnoses or direct medication doses.

Its safety rules prohibit adding a diagnosis of its own and telling a Client what amount of a medication to take or change. It may explain how a medication works and what to ask a prescriber, and it may discuss a diagnosis the clinician has already recorded, with clear attribution. The clinician can close either topic for a Client.

Crisis screening has no off switch.

Automatic safety screening and safety resources are independent of license payment. Only a crisis flag reaches clinical staff, as a review item and notification. Clinicians cannot read companion conversations, except the flagged message and the few around it that a crisis flag shows. Screening can miss risk; a notification is not proof that a person has read it. The Services are not a monitored crisis line and do not guarantee a response time. For immediate help in the United States, call or text 988; call 911 for an emergency.

The Services do not optimize for engagement.

The Companion has no streaks or engagement badges and is not designed to maximize conversation time. Usage counts support service operation and license administration.

Clinical decisions are made by the clinician.

The Companion supports the clinician’s approved care direction between sessions. It does not independently change treatment. The Services do not dispatch emergency services.

A separate AI prepares clinical observations for the care team.

A separate AI model reviews each Client’s companion conversations, including messages the Client did not save, and prepares structured observations for the clinicians and supervisors named in the Client’s sharing agreement. Observations are for clinician review; they are not diagnoses and are not automatically sent to the Client. Agreeing to the companion includes this review, and there is no separate switch for it. Future review stops when the practice removes the Client’s companion access or the Client closes their account.

The Companion remembers across conversations.

The Companion keeps memories that carry context between conversations. Clients can review, fix or forget them and turn automatic memories off. Clinicians cannot open private memory.

Clinician AI tools draft; the clinician decides.

Clinicians can use AI to draft briefs, plans and notes and to ask questions about a Client’s chart. These tools do not read companion conversations or private memory. Their output is an editable draft, and the clinician decides what to use.

AI processing supports the requested care workflow.

AI processing supports companion replies, safety screening, clinical observations and clinician drafting tools, each within the current care relationship and its permissions. The product is not a continuously acting autonomous clinician.

Infrastructure figures are attributed to their source.

The evidence and citations pages identify the infrastructure operator’s published energy and water figures. These describe the operator’s reporting boundary, not a measured footprint for the Services or an individual conversation.

Environmental claims require measurements.

Mojo AI Services LLC does not claim that the Services have zero emissions or no water use. This page does not claim a verified emissions reduction or carbon-removal contribution.