# ai-assistants

> Customer-facing AI Assistants: business-scoped, disclosed, with human handoff.

Customer-facing AI Assistants: business-scoped, disclosed, with human handoff.

| Method | Path | Operation | Scope | Test mode |
|---|---|---|---|---|
| GET | `/v1/ai-assistants` | `ai_assistant_list`: List assistants | `ai:read` | - |
| POST | `/v1/ai-assistants` | `ai_assistant_create`: Create an assistant (several per workspace, plan-limited) | `ai:write` | - |
| DELETE | `/v1/ai-assistants/{assistant_id}` | `ai_assistant_delete`: Delete an assistant | `ai:write` | - |
| GET | `/v1/ai-assistants/{assistant_id}` | `ai_assistant_get`: Get an assistant | `ai:read` | - |
| PATCH | `/v1/ai-assistants/{assistant_id}` | `ai_assistants_update`: Update an assistant | `ai:write` | - |
| POST | `/v1/ai-assistants/{assistant_id}/activate` | `ai_assistant_activate`: Go live (readiness checks) | `ai:write` | - |
| GET | `/v1/ai-assistants/{assistant_id}/analytics` | `analytics_ai_assistant_report`: Resolution, handoffs, credits, thumbs, unanswered | `analytics:read` | - |
| GET | `/v1/ai-assistants/{assistant_id}/benchmarks` | `ai_assistant_benchmarks_list`: Per-language scorecards over time | `ai:read` | - |
| GET | `/v1/ai-assistants/{assistant_id}/benchmarks/{run_id}` | `ai_assistant_benchmarks_get`: One per-language scorecard with every item | `ai:read` | - |
| POST | `/v1/ai-assistants/{assistant_id}/benchmarks/run` | `ai_assistant_benchmark_run`: Run the per-language quality scoreboard (queued) | `ai:write` | - |
| GET | `/v1/ai-assistants/{assistant_id}/conversations` | `ai_assistant_conversations_list`: AI-handled chats with resolution/handoff | `ai:read` | - |
| POST | `/v1/ai-assistants/{assistant_id}/deactivate` | `ai_assistant_pause`: Turn off (kill switch) | `ai:write` | - |
| GET | `/v1/ai-assistants/{assistant_id}/evals` | `ai_assistant_evals_list`: Eval results | `ai:read` | - |
| GET | `/v1/ai-assistants/{assistant_id}/evals/{run_id}` | `ai_assistant_evals_get`: One eval run with per-scenario verdicts, transcripts and the diff | `ai:read` | - |
| POST | `/v1/ai-assistants/{assistant_id}/evals/run` | `ai_assistant_evals_run`: Run golden conversations (queued) | `ai:write` | - |
| POST | `/v1/ai-assistants/{assistant_id}/feedback` | `ai_assistant_review_answer`: Thumb or flag an AI reply | `ai:write` | - |
| POST | `/v1/ai-assistants/{assistant_id}/publish` | `ai_assistant_publish`: Publish an assistant change behind the evaluation gate | `ai:write` | - |
| GET | `/v1/ai-assistants/{assistant_id}/runs` | `ai_assistant_runs_list`: AI runs with status, credits, reasons | `ai:read` | - |
| GET | `/v1/ai-assistants/{assistant_id}/scenarios` | `ai_assistant_scenarios_list`: Saved evaluation scenarios | `ai:read` | - |
| POST | `/v1/ai-assistants/{assistant_id}/scenarios` | `ai_assistant_scenarios_create`: Save an evaluation scenario | `ai:write` | - |
| POST | `/v1/ai-assistants/{assistant_id}/scenarios:generate` | `ai_assistant_scenarios_generate`: Generate scenarios from the knowledge base or a real conversation | `ai:write` | - |
| DELETE | `/v1/ai-assistants/{assistant_id}/scenarios/{scenario_id}` | `ai_assistant_scenarios_delete`: Delete an evaluation scenario | `ai:write` | - |
| GET | `/v1/ai-assistants/{assistant_id}/scenarios/{scenario_id}` | `ai_assistant_scenarios_get`: Get one evaluation scenario | `ai:read` | - |
| PATCH | `/v1/ai-assistants/{assistant_id}/scenarios/{scenario_id}` | `ai_assistant_scenarios_update`: Update an evaluation scenario | `ai:write` | - |
| POST | `/v1/ai-assistants/{assistant_id}/test-chat` | `ai_assistant_test_reply`: Test chat (same harness, tools dry-run) | `ai:read` | - |
| GET | `/v1/ai-assistants/{assistant_id}/test-chat/{session_id}` | `ai_assistant_test_sessions_get`: Replay a test session | `ai:read` | - |
| GET | `/v1/ai-eval-settings` | `ai_eval_settings_get`: Pre-publish gate and eval budget | `ai:read` | - |
| PUT | `/v1/ai-eval-settings` | `ai_eval_settings_update`: Update the pre-publish gate and eval budget | `ai:write` | - |
| GET | `/v1/ai-quality/report` | `ai_quality_report`: Quality by assistant, by agent or over time, worst first | `analytics:read` | - |
| GET | `/v1/ai-quality/scores` | `ai_quality_scores_list`: Conversation quality scores | `ai:read` | - |
| POST | `/v1/ai-quality/scores` | `ai_quality_score_request`: Score one conversation now (queued) | `ai:write` | - |
| GET | `/v1/ai-quality/settings` | `ai_quality_settings_get`: Conversation quality rubric and sampling | `ai:read` | - |
| PUT | `/v1/ai-quality/settings` | `ai_quality_settings_update`: Update conversation quality scoring | `ai:write` | - |
| GET | `/v1/ai-usage` | `ai_usage_get`: AI credit consumption by feature | `billing:read` | - |

Request and response schemas: the interactive reference at `https://api.wapitick.com/docs`, or `docs/api/openapi.json`.
