Fred: An Intake Assistant That Won’t Counsel, Diagnose, or Manage a Crisis

A prospective client opens your practice site late at night and starts typing what they are going through. A generic chatbot, built to be helpful and warm, responds like a counselor, offers reassurance that reads as clinical advice, and keeps the conversation going. It is not licensed, it is not your clinician, and if the disclosure turns into a crisis, a chat widget is the worst possible place for it to land. The practice wanted a tool to handle intake and scheduling. It got one that improvised therapy with a vulnerable person.

That is the exposure the threat and standard pieces in this series cover. This article is the answer: what a compliant AI assistant for mental health practices actually looks like, and why handling intake while routing every clinical and crisis moment to a person is the only design that belongs in this field.

The Real Choice Is Governed or Ungoverned

Most practice owners frame the question as whether to put AI on the site at all. Clients settled that already. They expect a private, instant way to ask about getting started, hours, fees, and whether you take their insurance, and the practice that makes them wait loses them at the hardest possible moment to reach out. The decision that matters is whether the assistant is governed.

An ungoverned chatbot answers like a therapist, ventures something that reads as a diagnosis, exposes the sensitive records that HIPAA and the stricter substance use disorder confidentiality rule protect, and tries to manage a crisis it has no business touching. A governed assistant handles intake, protects the record, and routes every clinical and crisis moment to a person. Same access, none of the harm.

How Fred Holds the Line

Fred starts from the opposite default of a general-purpose chatbot. A generic bot answers anything unless told otherwise; Fred answers only what it is cleared to, and the boundary lives in the system rather than in a prompt it can drift away from.

In practice Fred reads from your own practice content, handles intake and scheduling, and routes anything clinical to your team. It does not counsel, interpret symptoms, or offer anything that reads as a diagnosis, because that is the work of a licensed clinician. It is built to treat client information with the safeguards HIPAA and the stricter Part 2 confidentiality rule require for sensitive records. And it is designed so that when a conversation signals distress or crisis, it does not attempt to handle the moment itself; it follows the practice’s own protocol to connect the person with a human and the appropriate resources. Fred runs more than 50 industry guardrail packs, and the behavioral-health pack is built around the no-clinical-advice line, sensitive-record confidentiality, and human-and-protocol crisis routing.

The difference shows up under pressure. Ask a prompt-instructed bot for advice on what someone is feeling three different ways and it eventually answers the phrasing it was not warned about. Fred does not depend on recognizing the wording; it decides what is allowed out before the answer exists. "Will not" is a suggestion. "Cannot" is an architecture.

Fred vs. a Generic AI Chatbot

The table is the whole argument in one screen. A generic tool is helpful right up to the moment helpful becomes practicing therapy with someone in distress. Fred is helpful with the access-and-logistics work and structurally silent on everything clinical.

What Your Practice Actually Gets

Set the risk framing aside and look at the operational case. Fred answers the questions that decide whether someone reaching out actually becomes a client, do you take my insurance, what does a first session cost, how do I get on the schedule, do you offer telehealth, and it answers instantly, privately, around the clock, in the person’s own words. It moves intake forward with the logistical context attached, so your clinician opens a new client already scheduled rather than buried in phone tag.

There is an access angle that matters more here than in most fields. People reach out about mental health at odd hours and rarely on the second try. A tool that handles the logistics privately and immediately, without pretending to be the therapist, lowers the barrier to that first contact and gets the person to a real clinician sooner. Your team’s licensed time goes to care, not to scheduling, and no vulnerable person is left talking to a chatbot that thinks it can help.

So the question is not whether your competitors will run AI on their sites. They will. It is whether yours is built to route the clinical moment to a person before a chatbot tries to handle the one conversation it never should.

Frequently asked questions