feat(search): add semantic similar session matching via Voyage AI embeddings

Adds vector-based similar session discovery using the existing Voyage AI
embedding infrastructure and pgvector cosine similarity search.

- New AISessionEmbedding model with vector(1024) column
- session_embedding_service: generate + upsert embeddings, find similar sessions
- Embeddings generated on session create (from problem_summary/domain) and
  updated on resolve (adds resolution_summary)
- GET /ai-sessions/{id}/similar endpoint returns top-N similar sessions
- Migration a7c9e3b1f402 creates ai_session_embeddings table

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

backend/app/models/__init__.py

@@ -48,6 +48,7 @@ from .notification_log import NotificationLog
 from .notification import Notification
 from .psa_activity_log import PsaActivityLog
 from .file_upload import FileUpload
+from .ai_session_embedding import AISessionEmbedding
 
 __all__ = [
     "User",
@@ -110,4 +111,5 @@ __all__ = [
     "Notification",
     "PsaActivityLog",
     "FileUpload",
+    "AISessionEmbedding",
 ]

backend/app/models/ai_session_embedding.py

@@ -0,0 +1,53 @@
+"""AI session embedding storage for similar-session matching.
+
+Stores vector embeddings of AI session content (problem summary, resolution,
+domain) for cosine similarity search via pgvector. One embedding per session.
+"""
+import uuid
+from datetime import datetime, timezone
+
+from sqlalchemy import String, Text, DateTime, ForeignKey, Index
+from sqlalchemy.orm import Mapped, mapped_column
+from sqlalchemy.dialects.postgresql import UUID
+
+from app.core.database import Base
+
+try:
+    from pgvector.sqlalchemy import Vector
+except ImportError:
+    Vector = None
+
+
+class AISessionEmbedding(Base):
+    __tablename__ = "ai_session_embeddings"
+    __table_args__ = (
+        Index("ix_ai_session_embeddings_account_id", "account_id"),
+        Index("ix_ai_session_embeddings_session_id", "session_id", unique=True),
+    )
+
+    id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True), primary_key=True, default=uuid.uuid4
+    )
+    session_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ai_sessions.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+    account_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("accounts.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+    chunk_text: Mapped[str] = mapped_column(Text, nullable=False)
+    embedding_model: Mapped[str] = mapped_column(
+        String(50), nullable=False, default="voyage-3.5"
+    )
+    # The embedding column is created via migration with vector(1024) type
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), default=lambda: datetime.now(timezone.utc)
+    )
+    updated_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True),
+        default=lambda: datetime.now(timezone.utc),
+        onupdate=lambda: datetime.now(timezone.utc),
+    )