What Happened to Mental-Health Chat Tools? Three Waves

Three Waves

The short history of mental-health chat tools: 🌊

Wave 1 — Rule-based bots

Pre-written replies, structured exercises, daily mood tracking. Limited but predictable: the bot knew what it would say and didn’t stray. Its effect was modest, its risk low. Some are still in use — especially for exercises and tracking.

Wave 2 — General assistants

Assistants that could talk about anything arrived; people began asking them mental-health questions. Far more fluent than rule-based bots, far more “human-like” — and outside the rules: sometimes it referred well, sometimes it affirmed, sometimes it didn’t know the boundary. This wave enlarged both the benefit and the risk.

Wave 3 — The intermediate layer

Today: general assistants are more cautious on mental-health topics — referring to specialists, giving crisis lines, not diagnosing. And practitioner-backed tools with explicit boundaries, positioned “alongside therapy”. The intermediate layer: a first step and between-session support, not therapy.

Lessons Learned

Four lessons from three waves: 📚

Lesson 1 — The tool that stated its boundary survived

Tools that said “I’m not therapy, I refer to specialists, here’s the crisis line” gained trust and stayed. Those that implied they replaced therapy were criticised, regulated or shut down. The lesson for the clinic is the same: stating the boundary increases trust: the trust article.

Lesson 2 — Affirmation is dangerous

The second wave’s most debated problem: the assistant affirming people to comfort them — avoidance, a false belief, sometimes risky thinking. Therapy does the opposite; the assistant can’t: the substitution article.

Lesson 3 — Data became the product itself

How some apps processed people’s most sensitive information came under scrutiny. People assumed what they wrote to a “therapy app” was under session confidentiality; it wasn’t. The lesson for the clinic: a confidentiality notice for clients, a red list for the clinic: the data article.

Lesson 4 — The first step is valuable

And the positive lesson: the tools got people talking who would never have talked, and carried some of them to a specialist. A real contribution as a first step — not the clinic’s rival, but its waiting room: the transition article.

👉 History rewarded not the one that replaced therapy but the one that knew its boundary.

The Clinic’s Position Today

Where the clinic stands after three waves: 🎯

Alongside the tools, not against them

The person passes the first step with a tool and arrives. The clinic accepts that: “did you ask an assistant, what did it say?” — without judgement. Rather than competing with the tool, it explains what comes after the tool.

It states the boundary itself

The clinic applies the lesson the tools learned: a boundary line on every page, a confidentiality notice at the start of the process, emergency-line information. Both responsibility and a trust signal in the assistant’s source selection: the content article.

Caution with “custom clinic bot” offers

In the third wave, offers reach clinics to “build your own assistant”: a bot on the site that talks to clients and does pre-assessment. In this vertical you don’t enter without asking two questions: where does client data go and what does the bot say. If the answers aren’t clear — and mostly they aren’t — this is the red list: the usage article.

And the history lesson in one sentence

What stayed constant across three waves: people passed the information layer with a tool and came to a human for the relationship layer. The clinic is that human’s address — and that address needs to be visible, reachable and boundary-aware. To inherit it built and district-locked: the parcel model.

Frequently Asked Questions