No psychometric profile available
This issue occurs when SotsAI cannot apply behavioral reasoning because a required psychometric profile is missing.
This is not an error in SotsAI.
It is a design constraint.
What this means
Section titled “What this means”SotsAI requires at least one psychometric profile (the user) to operate.
If no user profile is available:
- SotsAI must not be called
- behavioral reasoning cannot be produced
- any attempt to do so will fail or be blocked upstream
This is intentional and non-negotiable.
Typical symptoms
Section titled “Typical symptoms”You may observe one or more of the following:
- your system refuses to call
/v1/advice - an error like
PROFILE_NOT_FOUNDorNO_VALID_ASSESSMENT - your orchestration layer skips SotsAI entirely
- guidance falls back to generic LLM behavior
Why this happens
Section titled “Why this happens”The most common causes are:
1) No profile was ever collected
Section titled “1) No profile was ever collected”- the user was never invited to complete a DISC assessment
- your system assumes profiles exist by default
- onboarding skipped the psychometric step
2) The profile exists but is not fetched
Section titled “2) The profile exists but is not fetched”- the DISC profile exists in SotsAI, but:
- you did not call
/v1/disc/profile - the result was not cached or reused
- you did not call
- your pipeline expects profiles to be “magically available”
SotsAI is stateless.
Profiles are never implicitly loaded.
3) The profile was fetched but not injected
Section titled “3) The profile was fetched but not injected”- the profile was fetched successfully
- but not included in the
/v1/advicepayload - or filtered out by schema validation or serialization logic
This often happens when:
- different profile schemas coexist
- optional fields are dropped silently
- tool payloads are rebuilt incorrectly
4) The tool was exposed too early
Section titled “4) The tool was exposed too early”- the LLM was allowed to call SotsAI
- before the profile resolution step completed
- resulting in blocked or skipped calls
This is a pipeline ordering issue.
How to fix it
Section titled “How to fix it”Choose one explicit strategy. Do not rely on implicit behavior.
Option 1: Gate the feature (strict)
Section titled “Option 1: Gate the feature (strict)”Recommended when personalization is core to the feature.
Flow:
- Detect missing user profile
- Do not expose the SotsAI tool
- Ask the user to complete a psychometric assessment
- Retry once the profile exists
Best for:
- coaching tools
- leadership support
- sensitive conversations
Option 2: Collect the profile asynchronously
Section titled “Option 2: Collect the profile asynchronously”Flow:
- Detect missing profile
- Trigger a DISC invitation
- Degrade gracefully for now
- Retry later when the profile is available
Best for:
- onboarding flows
- copilots with memory
- background enrichment
Option 3: Skip SotsAI (generic fallback)
Section titled “Option 3: Skip SotsAI (generic fallback)”Flow:
- Detect missing profile
- Do not call SotsAI
- Let the LLM handle the request autonomously
Best for:
- low-risk interactions
- optional personalization
- early experiments
What NOT to do
Section titled “What NOT to do”Avoid these anti-patterns:
- ❌ guessing or simulating a profile
- ❌ inferring traits from user text
- ❌ calling SotsAI “just in case”
- ❌ letting the LLM invent psychometric data
These break:
- behavioral validity
- user trust
- compliance expectations
Checklist for debugging
Section titled “Checklist for debugging”Before calling /v1/advice, verify:
- user psychometric profile exists
- profile was fetched or loaded
- profile is injected into the request payload
- tool is exposed after profile resolution
- no schema silently drops profile fields
If any box is unchecked, SotsAI should not be called.
Key takeaway
Section titled “Key takeaway”No profile is not an error. It’s a decision point.
Your system must decide:
- to block
- to collect
- or to fall back
Once that decision is explicit, this issue disappears entirely.