Skip to content

Profile strategy

SotsAI’s behavioral reasoning depends on psychometric profiles.

This page explains:

  • what counts as a valid profile
  • how to provide profiles
  • what to do when profiles are missing
  • how to design a sane fallback strategy

SotsAI requires at least a user psychometric profile to operate.

If no user profile is available, your orchestration layer must skip the SotsAI call and let your LLM handle the request autonomously.


A profile is structured data describing communication and decision preferences.

It does not need to be:

  • a personality diagnosis
  • a full HR report
  • DISC specifically

It does need to:

  • represent communication and decision preferences
  • be expressed as structured, explicit data
  • come from a known psychometric source or assessment
  • be stable enough to reuse across situations

SotsAI does not infer profiles from text, conversation history, or behavior. Profiles must be provided explicitly.


If you already have psychometric data, this is the best option.

Examples:

  • DISC
  • MBTI
  • Insights Discovery

You map your internal representation into the profile schema expected by SotsAI.

Pros

  • no additional user flow
  • immediate value
  • no credits required

Cons

  • requires internal alignment and mapping

You are responsible for ensuring the profile reflects real assessment data, not inferred or simulated traits.

If you do not have psychometric data, SotsAI provides a built-in DISC flow.

This allows you to:

  • invite users via email
  • collect profiles securely
  • fetch profiles on demand via API

Use this when:

  • no psychometric data exists
  • existing tools cannot be reused safely
  • you want an API-first, lightweight approach

DISC credits are only required to create profiles. Once collected, profiles can be reused freely.

Email addresses are required only to interact with the DISC assessment provider. SotsAI does not persist profiles or expose emails in advice calls.

Many teams use a hybrid approach:

  • use existing profiles when available
  • fall back to SotsAI DISC when missing
  • reuse profiles once collected

This minimizes friction while preserving behavioral quality.


Call SotsAI with:

  • context_summary
  • user_profile

This enables:

  • self-adjustment
  • reflection
  • better framing of messages

Useful when:

  • the interlocutor is unknown
  • the situation is one-sided
  • you want guidance focused on the user’s behavior

If no user profile is available, this call must not be made.

Call SotsAI with:

  • context_summary
  • user_profile
  • interlocutor_profile

This enables:

  • friction analysis
  • perception gap detection
  • adaptation strategies
  • tone and pacing guidance

This is the intended and highest-value usage.


If a user psychometric profile is missing, you must not call SotsAI.

You have two valid strategies.

Section titled “Option 1: Gate the experience (recommended)”

If the user profile is missing:

  • do not expose the SotsAI tool
  • inform the user that personalization requires a profile
  • offer a way to complete an assessment

Use this when:

  • behavioral accuracy is part of the product promise
  • advice quality matters
  • trust and consistency are critical

If the user profile is missing:

  1. trigger a DISC invitation
  2. notify the user
  3. handle the current request without SotsAI
  4. retry future requests once the profile exists

Avoid these anti-patterns:

  • calling SotsAI without a user psychometric profile
  • guessing profiles
  • inferring traits from a single message
  • embedding psychometric logic in prompts
  • forcing the LLM to “simulate” personalities

This leads to:

  • unreliable output
  • ethical issues
  • loss of trust

Profiles are generally stable.

You do not need to:

  • refresh profiles per request
  • update them per conversation

Update profiles only when:

  • the underlying assessment changes
  • your organization decides to re-run assessments

User psychometric profiles are a hard requirement for SotsAI’s behavioral reasoning.

Your integration should:

  • treat profiles as first-class data
  • gate SotsAI calls explicitly
  • handle missing profiles intentionally
  • never rely on inference or simulation