The chatbot is now part of the mental health workflow

Patients are turning to chatbots before therapy and our mental healthcare systems aren't ready for it, writes Alexander Amatus.

THE CHATBOT is now part of the mental health workflow, whether services planned for it or not.

In mental health, we tend to think the “front door” is the GP referral, the online booking form, the intake call, or the first session.

For a growing number of people, it is none of those.

It is a chatbot.

Not because people think a chatbot is a psychologist. Most don’t. It is because chatbots are available at 11:40pm, they don’t interrupt, they help people organise their thoughts and they lower the emotional friction of asking for help. For someone who is overwhelmed, embarrassed, time-poor, or not ready to talk to a real person yet, that matters.

The problem is not that this is happening. The problem is that many services still behave as if it isn’t.

That gap matters for health IT teams, digital leads, practice owners and clinicians, because the chatbot is now influencing what reaches your systems, your staff and your clinicians before the patient ever enters formal care.

Australia’s AI paradox: Mass adoption, minimal strategy

As generative AI becomes an everyday tool for Australians, new OECD data exposes a dangerous gap between mass uptake and the absence of any serious national strategy to govern it.

The new reality: “AI-shaped” patients are already arriving

Across mental health services, we are seeing a pattern where:

people arrive with AI-generated symptom summaries
they use chatbots to draft messages to clinics or employers
they paste chatbot language into intake forms
they ask for a specific diagnosis because “the bot said it could be…”
they use AI between sessions to journal, rehearse conversations, or seek reassurance

Some of this is genuinely helpful. In fact, some patients present better because they have done the hard work of putting chaotic thoughts into words.

But some of it creates new risks, like:

false certainty (“I already know what this is”)
reassurance loops (repeatedly checking the same fear)
delayed help-seeking (using AI as a substitute for booking)
privacy drift (sharing sensitive personal information into tools they don’t understand)
workflow friction (staff and clinicians spending time untangling what was patient voice vs generated text)

This is not a fringe issue. It is now a practical operations issue.

Why this is a health IT issue, not just a clinical one

Mental health services often treat AI use as a clinical conversation only. It is that, but it is also a systems and governance conversation.

The moment patients start using chatbots before care, the effects show up in:

digital intake design
triage scripts
consent and privacy language
documentation workflows
staff training
risk escalation pathways
patient communications

In other words, this lands squarely in health IT and service design.

If your service has upgraded telehealth, patient messaging, online bookings and digital forms over the last few years, this is the next layer. Patients are now arriving with a consumer AI layer on top of your official workflow.

You can ignore it, or you can design for it.

What good services are starting to do

The most useful response is not to ban AI, and it is not to promote it blindly. It is to set clear boundaries and build simple, human-centred pathways around what patients are already doing.

Here are five practical moves mental health providers and health IT leaders can make now.

1) Update the “digital front door” language

Most websites and booking forms still say nothing about AI. That leaves patients guessing.

A short, clear message can do a lot of work, for example:

It is okay to bring notes or summaries you have prepared (including AI-assisted ones)
AI tools can help organise thoughts
AI tools should not be used for diagnosis, emergency advice or crisis support
If there is immediate risk, contact crisis services or emergency care

This reduces shame, lowers confusion, and helps patients arrive with better expectations. It also gives your admin team and clinicians a consistent line to use.

2) Build AI-aware triage scripts for non-clinical staff

Reception and intake staff are often the first people to hear: “I put this into ChatGPT and it said…”

Without guidance, staff can get pulled into clinical conversations or dismiss the patient outright.

Neither response helps.

A better approach is a simple script framework:

Acknowledge: “Thanks for sharing that”
Reframe: “Online tools can be useful for organising thoughts, but they can’t assess risk or diagnose”
Redirect: “Let’s get you booked with the right clinician/triage pathway”
Escalate if needed: follow existing risk protocols

This is basic operational hygiene. It protects staff, supports patients and keeps role boundaries intact.

3) Treat AI-generated text like any other patient-supplied material

Some services are unsure what to do with AI-generated summaries in records.

The answer is usually straightforward: treat them as patient-provided information, not clinical conclusions.

That means one should:

note the source where relevant (for example, patient-provided summary)
verify key details clinically
avoid copy-pasting generated content into notes without review
separate patient narrative from clinician formulation

This is not just documentation quality. It matters for medico-legal clarity and continuity of care.

If a service is serious about governance, this point should be explicit in internal documentation standards.

4) Add privacy warnings where patients are most likely to need them

Most privacy notices are written for lawyers, not patients. They are also usually placed in the wrong spot.

If patients are using AI to prepare for care, the key message should appear at moments of behaviour:

on intake forms
in booking confirmations
in pre-appointment emails
in FAQs
in telehealth onboarding pages

Keep it plain:

avoid entering highly sensitive identifying information into public AI tools
use AI for drafting and organising, not for emergencies
bring questions to your clinician

This is one of the easiest wins in digital health communication right now.

Caring as a revolutionary act: The fight for humane mental health practice

Our clinicians are navigating mental health systems that are deeply broken — systems that too often fall short of offering dignity or true healing to those needing help. Yet, there is hope.

5) Stop treating this as a one-team problem

AI in mental health workflows touches:

clinical governance
privacy/compliance
digital/product
practice operations
workforce training

If only one team is handling it, you will get blind spots.

The services that will handle this well are the ones that create a small cross-functional working group and decide a few basics quickly, such as:

what patients can be told
what staff can say
what is documented
what is escalated
what tools (if any) are approved internally
what is explicitly not allowed

This does not need a 40-page policy to start. It needs a usable one-page protocol and staff training.

The bigger opportunity: AI can reduce friction if we design the guardrails

There is a tendency to frame this as either a threat or a revolution.

In practice, it is more ordinary than that.

Patients are using AI because mental health access is hard. Booking is hard. Naming symptoms is hard. Asking for help is hard. Explaining your story for the fifth time is hard.

If AI helps someone get from “I’m spiralling” to “I’ve booked an appointment and written down what I need to say,” that is useful.

If it keeps them stuck in a loop, delays care, or creates bad certainty, it is harmful.

The difference is often not the tool. It is the guardrails around the tool.

That is why this is such an important moment for health IT in mental health. The task is not to predict the future. The task is to design workflows that match the reality patients are already living in.

The chatbot is already in the pathway.

The question now is whether the rest of the system is ready for that.

Alexander Amatus works at the intersection of AI and mental health service leadership at TherapyNearMe.com.au, Australia’s fastest-growing mental health service, with a focus on practical, human-centred integration of digital tools into real-world care pathways.