Statutory conversation practice — complete practitioner guide

Default home is Start conversation. Choose a situation or More options (Lab) if you want themes or the card deck. In Lab, a tiny chip toggles Quick start vs Custom. Both still send your choices to the AI — there is no offline case library.

FAQ: Quick start or Custom · Adult vs Children's themes.

Every KallosSim page has Sign in / Account in the header. Hub Start free opens Create account. Free practice needs a real account first — Create account or Sign in — then pick your role and start a conversation. Quota sits on the account, not an anonymous cookie. On production (app.kallossim.com) use Continue with Google, or email and password. The same Google address as an existing password account opens that account — you do not get a second login. Pilot invites must be accepted with the invited email.

A Vercel git preview (*.vercel.app, except dialogtrainer-en.vercel.app) signs you in as QA Guest and does not use Google. Production and that Vercel production alias reject QA Guest. If you are on a preview, stay on that host — in-app Guide and Back to app stay on the same deployment. Public “Open app” links on the marketing site go to production; if a link jumps environments, use Back or paste your preview URL. Full answers: FAQ — sign in · preview or production · demo links.

Pricing and the app use Free, Individual, Company, and Larger organisation. A Stripe receipt may show an internal id: professional = Individual, team = Company (5 seats), organisation = Larger organisation. FAQ: plan names.

Invited testers can install KallosSim from Apple TestFlight. This is internal testing, not an App Store listing. Short shareable page: /ios. FAQ: iPhone TestFlight.

Install: get TestFlight from the App Store → open the invite you were sent (same Apple ID) → Install → sign in with the same email and password as the website.

Internal TestFlight currently installs KallosSim (Capacitor) com.kallossim.app. KallosSim Native com.kallossim.app.native is a Phase 1 sibling — it may be installed separately until it shares the TestFlight listing. Native is Sign in, Start conversation, and text chat — not the card deck, voice, debrief, Google sign-in, or other sectors. Tell them apart by the name under the icon; both source icons use the same brand mark.

There is currently no in-app Help (no ? / Hjelp) on iOS Native. Use this handbook and the FAQ in Safari or Chrome.

Legal Advocacy practises oral advocacy steps (opening, examination-in-chief, cross-examination, procedural objection, closing). It is a learning tool for England & Wales courtroom skills — not a Bar Standards Board, SRA, or Inn competence assessment. Cases and personas are AI-generated. Enter via /legal-advocacy/app.

Registered Intermediaries have a primary duty to the court, not to a local authority safeguarding process. Card B process stages are criminal justice steps — intermediary request, communication assessment, ABE planning, ABE interview support, Ground Rules Hearing, and criminal trial evidence. The AI knowledge pack is criminal_justice, not Working Together. RI practitioners should enter via /witness-intermediary/app, not Children's Services.

Registered Manager (Children's Home) — Ofsted children's residential; appears in Children's Services. Registered Manager (Adult Care) — CQC adult care home; appears only in Adult Social Care. They share a label in conversation but have different statutory frameworks, process stages, and evaluation emphasis.

Your role (Card A) filters Cards B (process stage), D (target group), E (dimension), F (conversation type), and I (training focus). Card F adapts when you change Card B. Card J (Maturity) appears only for young-person target groups. Shuffle respects these constraints — you will not land on an incoherent combination.

On the home page, services, and pricing you may see a banner such as Practising in Canada? Our Ontario · CYFSA 2017 training pack may fit better than UK Working Together. Choose View Canada pack to open the landing page, Stay on UK to dismiss for this visit, or × to hide suggestions for 30 days. See privacy — cookies & local storage.

International pilots are children's services only. Adult social care, health, youth justice, witness intermediary, HR, and Legal Advocacy remain UK-only. Full reference: international markets user help (Markdown).

Card A is the most important selection you make. Your professional role is chosen on the role picker before you enter the deck, then shown on the role banner at the top. Your role determines which process stages, target groups, and training focuses appear on the remaining cards, and how the AI persona perceives your authority. A parent wary of a statutory Social Worker will respond very differently to a DSL from their child's school — or to a Registered Intermediary in a court setting.

Registered Manager (Children's Home) is Ofsted children's residential. CQC adult Registered Manager is only in Adult Social Care.

These labels are easy to mix up. The table below restates the in-repo children's knowledge pack (WT 2026 / FFP / Children Act 1989) — not a new statutory claim. Fag review via Boss if local practice uses different labels.

Glossary: Family Help pathway · Section 17 · Section 47 · SENCO. FAQ: Family Help, CAF, CFA.

Each of the remaining nine cards controls one dimension of the generated scenario. Click any card to flip it and choose from all available options, or shuffle freely.

After a case is generated, you may see an optional control: Was this scenario realistic? with 👍 and 👎 buttons. This is separate from your debrief score — it does not affect assessment.

KallosSim has built-in GDPR controls and organisation DPO support — not only a privacy page. On first login you see two notices: how learning data is stored, and how anonymised improvement data may be used. You can review or change preferences anytime from your profile (top-right menu).

Full practice conversations are not stored in our database — only scores, metadata, and short evaluation excerpts (up to 30 days). Contact form submissions are stored to respond and synced to our CRM queue. Two consent surfaces: improvement analytics (usage events, off unless you opt in) is not the same as research Yes/No in My data usage. The marketing-site Accept analytics banner is a third, page-view choice — also not research consent. FAQ: analytics vs research. Full legal detail: Privacy notice · User guide: Built-in GDPR & DPO · Technical design: feedback pipeline (for DPOs). Deployer duties: AI Act for organisations.

If your organisation assigns KallosSim to staff who practise with a simulated person, you are a deployer. That is different from KallosSim as the provider. You do not sign Article 6, affix CE, or file the EU AI database — those are provider decisions, and KallosSim currently claims neither CE marking nor an EU-database filing.

Full deployer note: transparency — deployers · In-app: Org Admin → Privacy & AI Act · Provider register stays in Editorial Admin (not shown to customers).

Before the conversation begins, review the case panel. It contains everything you need to orient yourself:

The in-app name is Practice advisor. It appears first after Generate succeeds — bottom-right, once a case is active. It is not on the empty Generator or the role picker. Ask it about this case. It is an AI mentor, not verified statutory doctrine, and not a substitute for supervision or legal advice. Legal Advocacy answers are general procedural themes — check citations independently. App or account questions stay with Support (bottom-left). FAQ: What is the Practice advisor?

Text is the default and does not use the voice allowance — use it to draft wording or save voice sessions. Voice lets you hear the persona and reply by microphone. Free includes 3 voice sessions in total, each capped at 10 minutes; then continue in text. Individual and Company include 10 voice sessions per month. Headphones help. FAQ: When should I use text or voice? · Free limits.

The conversation is structured into five phases. Each phase has a distinct purpose and a coaching prompt visible in the phase header. You do not have to complete all five — End & evaluate is always available.

The RC (Relational Capital) bar is a real-time trust score. It reflects how the persona's trust in you is moving during the conversation, derived from their minority stress profile and your communication choices.

When a scenario has statutory time pressure — for example, a Section 47 enquiry or an Emergency Protection Order — a clock indicator appears at the top of the conversation view. It shows:

Purpose: to practise the discipline of keeping legal deadlines in mind during emotionally pressured conversations — a core competency for statutory practitioners.

the AI evaluator reads your full conversation transcript and scores your communication against the criteria loaded for this sector and role (an upper bound of about 40 — not a fixed 5/27 scorecard). The evaluator is a separate AI system. It receives your full transcript, the scenario context, your professional role, and the training focus for this session — then scores each criterion on a 0–3 scale. For each criterion it produces: a numeric score, a written explanation of your strengths, and specific development suggestions with quotes from the conversation.

In addition to criterion scores, the debrief includes a BARS (Behaviourally Anchored Rating Scale) summary across four axes. These give a sector-level view of your practice independent of individual criterion scores.

MI (Motivational Interviewing) is a collaborative, person-centred communication style designed to strengthen a person's own motivation for change. It is evidence-based and widely used across social work, healthcare, addiction, and HR. KallosSim scores five MI skills using the OARS+E framework:

The Learning Circle generator takes your team's practice data and creates a 6-session structured programme. Each session plan includes:

Supervisors can see debrief scores, criterion-level breakdowns, and short AI-selected trainee excerpts in the rubric (not full session transcripts). Excerpts are retained for up to 30 days; full dialogue is never stored in the database. All scenario content is AI-generated — no real case data is ever stored. Practitioners are identified in the Editorial view only by the name they registered with.

Organisation administrators (team lead, senior manager, DPO) also have an Org Admin workspace separate from Editorial. Open it from the profile menu. The Quality tab shows:

Settings tab includes a DPO note on how improvement data is scoped to your organisation. Full DPIA status and org sign-off: Org Admin → Privacy & AI Act. See also Privacy & improvement for practitioner controls.