Clinician Tools

The Icosa clinician portal is a complete clinical workspace built around the same computational engine that powers every assessment. Client management, clinical documentation, session analysis, persona classification, behavioral signal detection, and multi-reporter consensus — everything you need to integrate structured personality data into therapeutic practice, without adding another hour of documentation to your week.

Built for How Clinicians Actually Work

Personality assessment tools tend to hand you a report and leave. Icosa gives you a living clinical workspace. From the moment a client accepts your invitation through years of ongoing therapeutic work, the clinician portal tracks, analyzes, and surfaces what matters — without adding another hour of documentation to your week.

Everything described here lives inside a dedicated Clinician SPA with 16 purpose-built screens. It is available at the Clinician Starter tier and above.

Client Management

The relationship between clinician and client begins with a simple email invitation. You send it from the portal; the client accepts from their personal dashboard. No new account creation, no onboarding friction — the link connects two existing Icosa profiles into a clinical relationship with appropriate access controls.

Once connected, you see every assessment the client has completed: their current formation, coherence band, behavioral signals, and longitudinal trajectory. The client controls whether you can see their clinical notes (three visibility levels ensure nothing is shared without consent). They can revoke your access at any time.

The portal supports up to 10 active clients on the Clinician Starter plan and unlimited clients on Clinician Pro.

Clinical Notes

Five note types cover the full range of clinical documentation: general, clinical, session, progress, and intake. Each note carries a visibility level — clinician-only (private to you), client-only (visible only to the client), or both (shared between clinician and client). You create, edit, and delete notes directly from the client detail screen.

Notes are stored with full HIPAA compliance: integrity hashing, audit logging, 7-year retention, and PHI scrubbing on all applicable outputs. They are included in GDPR data exports from both the clinician and client perspectives.

This is not a replacement for your EHR. It is a clinical annotation layer that lives alongside the assessment data, so your observations sit next to the evidence that prompted them.

Session Analysis

Paste up to 200,000 characters of therapy session transcript into the session analysis tool. The system scrubs all PII before processing, analyzes the text for Icosa-relevant patterns, and returns a structured breakdown of what the session reveals about the client’s current grid state.

The transcript is never stored. It is scrubbed, analyzed, and discarded. Only the analysis results persist — and those are saved to the client’s record for longitudinal tracking.

The analysis identifies which centers are activated in the session material, with three confidence tiers: direct (explicitly stated), implied (strongly suggested by context), and speculative (possible but not confirmed). This gives you a structured lens on unstructured clinical material.

The Session Analysis Pipeline

The pipeline follows four steps, each designed to protect privacy while extracting maximum clinical insight:

1. PII scrubbing — 9 pattern types (email, SSN, phone, dates, IP, UUID, MRN, ZIP, signature) are identified and removed before any LLM processing. Nothing identifiable reaches the AI model.

2. Grid mapping — The LLM maps natural language observations to specific grid states across the 20 centers, assigning confidence levels to each inference. The mapping uses the same structural vocabulary as the formal assessment, making session analysis results directly comparable to assessment profiles.

3. Visualization — Results render as an Icosaglyph with color-coded badges showing which centers were identified and at what confidence level. You can overlay this session-derived glyph on the client’s last formal assessment to spot convergences and divergences.

4. Storage — Analysis results (not transcripts) are saved and linked to the client record. Transcripts are discarded immediately after processing.

The entire flow runs under HIPAA mode: raw LLM responses are cleared after extraction, preventing PHI from persisting in cached or logged data.

Text Interpreter

Not every clinical observation comes from a structured session. The text interpreter takes free-form text — a paragraph of clinical notes, a brief observation, a client’s self-description — and maps it to grid states.

Type or paste any natural language description of a person’s functioning, and the system converts it into a full grid profile. Each inferred center comes with a confidence tier. The result includes an Icosaglyph visualization you can compare against the client’s actual assessment.

This is particularly useful for initial intake: before a client has taken a formal assessment, you can sketch a provisional profile from your clinical observations and compare it later against the assessment results.

SOAP Note Generation

The system generates Subjective, Objective, Assessment, Plan notes from assessment data using the Reports v2 engine with a SOAP preset. These are formatted for clinical use: copyable, exportable, and print-optimized.

SOAP notes pull from the full profile computation — 355+ data fields including coherence scores, gateway identification, trap detection, formation classification, and behavioral overlays. The result is a structured clinical note grounded in computed data rather than subjective recall.

Persona Classification

Every client profile is matched against 169 individual personas — named clinical archetypes with associated DSM codes, severity ratings, coherence bands, characteristic traps, and basins. The classification uses seven-dimension evaluation with six layers of progressive contextual refinement:

1. Structural — Grid geometry matching
2. Contextual — Demographic and life event context
3. Demographic — Age, gender, and background factors
4. Life event — Recent events and their impact zones
5. Behavioral — Assessment-taking behavior signals
6. Living situation — Family structure and environment

The result is a ranked list of matching personas with confidence scores, evidence bundles, and contradiction detection. When a client matches multiple personas, the system identifies which evidence supports and which contradicts each match.

What Persona Classification Adds to Treatment

Persona matching serves three clinical functions that go beyond labeling.

Treatment matching — Each persona carries clinical annotations suggesting which therapeutic approaches are most effective for that configuration. The annotations are grounded in the persona’s specific grid signature, not generic type-based recommendations.

Progress tracking — As a client’s profile shifts across assessments, their persona match may change. Moving from a Struggling-band persona to a Steady-band persona provides a qualitative marker of progress that complements the quantitative coherence shift. The shift itself is clinically informative: which persona are they moving toward, and does that align with therapeutic goals?

Cross-source corroboration — When self-assessment, clinician assessment, and behavioral signals all converge on the same persona, clinical confidence is high. When they diverge, the divergence itself is diagnostically informative — it reveals where self-perception, external observation, and behavioral data tell different stories.

Behavioral Signal Detection

During every assessment, the system captures 23 behavioral signals across 5 categories — invisible to the person taking the assessment but visible to you in the clinician portal.

Signal confidence scales with assessment tier: Quick (0.30), Standard (0.70), Comprehensive (1.00). Longer assessments produce more behavioral data and higher-confidence signals.

These signals serve as a validity check on the self-report data. A client who answers every question in under two seconds, shows high monotonic response patterns, and exhibits low trajectory variance may be rushing through the assessment. The behavioral layer flags this so you can interpret the profile accordingly — or request a retake.

Multi-Reporter Consensus

The clinician assessment is not a replacement for the client’s self-assessment — it is a second perspective. When both exist, the multi-reporter system computes a consensus profile using weighted multi-perspective integration across self-report, other-report, clinician observation, and population baseline.

The consensus view reveals blind spots: areas where the client’s self-perception diverges significantly from the clinician’s observation. The Self-Awareness Index (0-100) quantifies this divergence. A high SAI means the client sees themselves roughly as you see them. A low SAI means there are meaningful perception gaps — which are themselves clinically informative.

Side-by-Side Comparison

Enter any two signatures — a client’s current and previous assessment, two clients you want to compare, or a self-report and clinician report — and the portal renders both profiles side by side with a full differential analysis.

This is the fastest way to see what changed between sessions, how two clients differ despite similar presenting concerns, or where the clinician’s perception diverges from the client’s self-report.

What This Means in Practice

The goal is to give you structured, computed insight into your clients’ functioning without requiring you to become a psychometrician. The system does the computation. You do the clinical judgment.

Availability

The clinician portal is available at Clinician Starter (up to ten active clients) and Clinician Pro (unlimited clients). Both tiers include the full clinical workspace, session analysis, note generation, and persona classification. Multi-reporter consensus requires at least two completed assessments for the same client.