Talk it Through is the only between-session AI with full context of the work. It reads the modules you assigned, every lesson your client has finished, and every reflection they've written beside each one - and walks them through the tough bits in their own data, in their own language, in their own moment.
Phase-aware. Integrative across CBT, DBT, ACT, Psychodynamic, IPT. Crisis-flagged. Captured to your chart. Governed by you.
A generic chatbot doesn't know your client did the catastrophising exercise on Thursday, or that they wrote four lines beside the cognitive-defusion lesson at 11pm last Tuesday. Talk it Through does. The Guide opens with the full context of the work already loaded - and it shapes every response.
When your client opens a tough lesson on Sunday night and gets stuck, they don't sit with it for a week and forget it ever happened. They tap the Guide. The Guide knows which lesson they're on, what the psychoeducation block actually says, what the exercise asked them to do, and what they wrote about it last time.
It doesn't explain the concept generically. It explains their version of it - anchored to what they wrote, what they've already noticed in their own life, what's come up in mood entries this week. The conversation is genuinely continuous with the work.
And because every reflection they write is read by the Guide, the AI's response evolves alongside their understanding. The Sunday-night chat about catastrophising sounds different in week six than it did in week one - because the data underneath it is different.
Beyond the 276 evidence-based modules in Anchr's library, you can write your own - your framing, your exercises, your language. The Guide treats them no differently. When you hit Publish, an AI pass reads the lesson content and drafts the orientation metadata - trigger phrases your client might say that map to the module, contraindications, techniques it teaches, and which preset modules pair well. You review and edit before it goes live. So between sessions, when your client says something the module addresses, the Guide notices - and walks them through the concept using your wording from the lesson, not a generic AI's approximation of it.
This is the part no other between-session AI does: the modules you wrote, taught by the Guide in your clinical voice, anchored to your client's own reflections beside each lesson. Bespoke content, between-session continuity, no hand-off.
Tuesday morning, that conversation is on your chart - with the moment Sarah finally got the practical-vs-hypothetical split. You walk into Thursday's session already inside it.
Talk it Through doesn't lecture, doesn't rush, and doesn't shove a technique at the first opportunity. It moves through a four-stage arc - selecting modality and depth based on where your client is right now, not where the AI wishes they were.
What is your client bringing today - a sticky thought, a hard week, a 2am realisation, a stuck moment in this week's lesson? Different opening, different work.
The middle of the work. Reflective listening, gentle assumption-testing, and techniques drawn from your client's active modules - woven in without lecturing.
One layer below the surface story - patterns, the want underneath the complaint, the value the pain points toward. Earned, never forced.
Pulling threads together. Take-aways, lesson cross-references, and what to bring back into the next session with you.
The Guide does not push your client forward before they are ready, and does not pretend the conversation is over while they still have more to say. Stage-of-change calibration is enforced: no action-oriented techniques to a precontemplative client; no schema work in session two.
Underneath the arc sits a compressed catalogue of roughly 170 evidence-based techniques across six clinical modalities, each paired with a clinical reference and a voice-calibrating example. The Guide selects from the library based on phase, the function of what your client brings, their active modules, and what's worked in earlier chats.
Thought records, behavioural activation, schema work, exposure, panic protocols
Mindfulness, distress tolerance, emotion regulation, DEAR MAN, chain analysis
Defusion, values clarification, acceptance, committed action
Pattern recognition, transference, mentalization, attachment, inner child
Communication analysis, role disputes, grief & transitions, boundaries
Modern stressors - financial, geopolitical, relational loneliness, AI anxiety
What makes Talk it Through different from any chatbot a client could turn to: every conversation is gated on three structured outputs before it ends. These are the artefacts that reach your record - the things that turn a between-session conversation from invisible into clinically actionable.
What was the focus, which techniques were used, which resonated, which threads were left open. The moment-of-truth summary - a one-page artefact that captures the shape of the conversation.
A signed structured note - themes, techniques tried, homework framed, risk markers, affect observations - with verbatim quote anchors and confidence labels. The kind of note you'd write yourself if you'd had the hour.
An evolving clinical formulation that updates session-over-session. New evidence is integrated into the working frame - not appended as noise. The Guide tracks what's shifted, what's stable, and what's now contradicted.
Conversations crossing a clinical threshold - hopelessness language, ideation, scope-of-practice signals - surface in the portal immediately. The Guide pauses, refuses to "therapise" a crisis, connects to crisis services, and notifies you.
Multi-tier safety protocols run in every conversation. Crisis content is surfaced - never silently filtered - even at low confidence. The Guide pauses, refuses to provide reassurance or behavioural prescriptions for an active crisis, connects to crisis services (Lifeline, 13 11 14 / Beyond Blue, 1300 22 4636 in AU), and propagates risk markers into your portal immediately.
Item-9 of PHQ-9 endorsements, hopelessness language, plan/intent/means signals - these reach you the same day, with the verbatim phrase and the timestamp. Built to escalate to a human, not to please a user.
Talk it Through doesn't diagnose, doesn't prescribe, doesn't function as a therapist, and doesn't replace clinical judgment. Every conversation lives in the client record, and the client knows that going in. They're not talking to a generic chatbot - they're talking to their Anchr Guide, scaffolded by their work, visible to their therapist, governed by you.
Client-side AI consent is explicit and opt-in. PHI is gated. AU-hosted, Privacy Act 1988, RLS-gated end to end. Every claim cites the field it was grounded in - the 2:14am chat, the lesson note from April 18, defusion entry #14 - so you can audit exactly what each line was based on.
And the boundary holds: the Guide is built to be a clinical bridge, not a clinical replacement. If a client of yours stops seeing you tomorrow, their data follows them, not us.
Talk it Through is the alternative your clients deserve to the generic chatbot they're already using - and the depth of context you've never had access to before.
Anchr is a self-care tool used alongside professional care. Talk it Through does not diagnose, prescribe, or replace clinical judgment. The Guide functions under therapist oversight, with explicit client consent, and never outside the boundaries of the therapeutic relationship.