---
id: JNY-PFM-LLM-002
title: Support investigates an AI refusal
feature: FEAT-PFM-LLM-04
product_area: platform
module: llm
revision: 1
status: draft
product_intent: proposed
implementation_evidence: observed
supersedes: []
source_sha:
  services-platform: ad3f6a6582843008ed0f5a17b131878d15829af1
  react-console: baf6dae382d6f41dce976c7f848667ccc37e076f
actor: ACT-PFM-033
goal: Explain an AI refusal to a customer and establish whether the request can proceed now.
capabilities:
  - CAP-PFM-LLM-010
  - CAP-PFM-LLM-011
  - CAP-PFM-LLM-012
screens:
  - /hq/aiUsage
  - '/hq/tenants/{id}'
  - /hq/aiUsage/requests
  - '/hq/aiUsage/requests/{tenantId}/{invocationId}'
steps:
  - 'n': 1
    surface: /hq/aiUsage
    action: Find the organization on the AI Usage page.
    capability: CAP-PFM-LLM-010
    expected: 'The list supplies its usage position, cost and status.'
    provenance: test
    source: >-
      react-console:e2e/mock/hq-ai-usage.mock.spec.ts#shows the tiles, the trend, who needs attention and the
      organizations with their states
  - 'n': 2
    surface: '/hq/tenants/{id}'
    action: Open the organization’s AI Usage card.
    capability: CAP-PFM-LLM-011
    expected: 'The card supplies allowance figures, reset information, limits and recent history.'
    provenance: test
    source: >-
      react-console:e2e/mock/hq-ai-usage.mock.spec.ts#reads the limited organization: the warning band, the
      limits, the history and the refusals
  - 'n': 3
    surface: /hq/aiUsage/requests
    action: Open View Requests and select refused requests.
    capability: CAP-PFM-LLM-011
    expected: The directory filters to that organization without displaying captured bodies.
    provenance: test
    source: >-
      react-console:e2e/mock/hq-ai-usage.mock.spec.ts#asks for an organization first, then filters its
      requests without ever showing prompt text
  - 'n': 4
    surface: '/hq/aiUsage/requests/{tenantId}/{invocationId}'
    action: Open a refused request.
    capability: CAP-PFM-LLM-011
    expected: 'The detail supplies request metadata, attempts and its refusal snapshot.'
    provenance: test
    source: >-
      services-platform:modules/llm/tests/e2e/llm-usage.e2e.spec.js#answers one invocation with its attempts
      and the refusal snapshot of a refused one
  - 'n': 5
    surface: '/hq/aiUsage/requests/{tenantId}/{invocationId}'
    action: Read the explanation.
    capability: CAP-PFM-LLM-012
    expected: 'The panel presents the recorded reason, admission arithmetic and a current comparison.'
    provenance: test
    source: >-
      react-console:e2e/mock/hq-ai-usage.mock.spec.ts#a refused request explains itself in three blocks, and
      shows evidence statuses without content
alternates:
  - from_step: 4
    condition: The request failed after dispatch.
    path: Read the attempt timeline and verdicts.
    outcome: Support can distinguish a provider failure from an admission refusal.
    provenance: test
    source: 'react-console:e2e/mock/hq-ai-usage.mock.spec.ts#a failed request shows its attempts as a timeline'
failures:
  - from_step: 1
    condition: A usage section cannot load.
    path: That section presents its unavailable state independently.
    outcome: Support can continue with information that loaded.
    provenance: observed
    source: 'react-console:components/console/hq/ai-usage/ai-usage-tiles.tsx'
  - from_step: 1
    condition: A customer administrator tries an HQ read.
    path: HQ authorization rejects the request.
    outcome: The customer receives no HQ usage data.
    provenance: test
    source: >-
      services-platform:modules/llm/tests/e2e/llm-usage.e2e.spec.js#refuses a customer’s administrator,
      whatever they hold at their customer’s scope
rules_in_play:
  - BR-CAP-PFM-LLM-010-01
  - BR-CAP-PFM-LLM-010-03
  - BR-CAP-PFM-LLM-011-05
  - BR-CAP-PFM-LLM-011-06
  - BR-CAP-PFM-LLM-011-12
  - BR-CAP-PFM-LLM-012-02
  - BR-CAP-PFM-LLM-012-04
assertions:
  - id: AS-JNY-PFM-LLM-002-01
    statement: 'Agreement between the overview, organization card and request directory remains unproved.'
    provenance: unknown
    source: FND-PFM-LLM-027
  - id: AS-JNY-PFM-LLM-002-02
    statement: The directory excludes captured request and response bodies.
    provenance: test
    source: >-
      react-console:e2e/mock/hq-ai-usage.mock.spec.ts#asks for an organization first, then filters its
      requests without ever showing prompt text
  - id: AS-JNY-PFM-LLM-002-03
    statement: The explanation presents refusal figures and a current admission comparison.
    provenance: test
    source: >-
      react-console:e2e/mock/hq-ai-usage.mock.spec.ts#a refused request explains itself in three blocks, and
      shows evidence statuses without content
e2e_policy: required
e2e_reason: >-
  This path depends on agreement between several HQ screens. Mock browser coverage establishes their
  presentation; the live browser check does not compare their figures.
evidence:
  - >-
    react-console at baf6dae382d6f41dce976c7f848667ccc37e076f: app/(console)/hq/aiUsage/**;
    app/(console)/hq/tenants/[id]/page.tsx; components/console/hq/ai-usage/**;
    components/console/hq/tenant-ai-usage-card.tsx; e2e/mock/hq-ai-usage.mock.spec.ts;
    e2e/dev/hq-ai-usage.dev.spec.ts.
  - >-
    services-platform at ad3f6a6582843008ed0f5a17b131878d15829af1: docs/llm.md (The HQ surface, Explaining a
    refusal); modules/llm/tests/e2e/llm-usage.e2e.spec.js; docs/milestones/plans/PLAT-M1-plan.md (S05, S09,
    S15).
proved_by: []
---

# Support investigates an AI refusal

## Story

A customer asks why AI refused a request. Support starts with the organization’s usage position, locates a refused request and reads its explanation. The journey connects the HQ reporting and explanation capabilities. Comparing the same figures across the complete path remains an acceptance obligation.