feat: Slate & Ice Modern aesthetic redesign (#94)

* chore: update Google Fonts to Bricolage Grotesque, IBM Plex Sans, JetBrains Mono Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: update Tailwind config to Slate & Ice theme colors and fonts Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: update CSS variables and glass-card utilities for Slate & Ice theme - Replace all color variables with Slate & Ice palette - Add glass system vars (--glass-bg, --glass-blur, --shadow-float) - Replace legacy glass-card with new variable-driven glass classes - Add breatheGlow, bellWobble, slideDown, fadeInRight keyframes - Update font references to IBM Plex Sans and Bricolage Grotesque Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: recolor BrandLogo to cyan gradient, split BrandWordmark for gradient Flow text Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: update TopBar with glassmorphism backdrop and cyan accent styling Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: update Sidebar with glassmorphism backdrop Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add ambient atmosphere gradient orbs behind app shell Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: update QuickStats and SessionsPanel with glass-card styling Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add WeeklyCalendar, QuickActions, OpenSessions, RecentActivity dashboard components Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: redesign dashboard layout with calendar, open sessions, and glass-card panels New layout: greeting → calendar+actions → sessions+stats → activity Replaces old QuickStats and SessionsPanel with new dashboard components Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: replace remaining purple hex references with ice-cyan accent Sweep of hardcoded purple hex values (#818cf8, #6366f1) replaced with new cyan accent (#06b6d4) in QuickActions, RecentActivity, QuickLaunch, and SVG brand assets. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs: update CLAUDE.md branding and design system for Slate & Ice Modern Updated Last Updated date, branding section (fonts, colors, glass utilities, atmosphere orbs), component styling rules, and Design System section to reflect the new ice-cyan glassmorphism theme. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs: add Slate & Ice Modern design doc and implementation plan Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: redesign login page with Slate & Ice Modern design system Apply glassmorphism styling, atmosphere orbs, branded wordmark, and consistent design tokens to match the updated app shell aesthetic. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: raise TopBar z-index so profile dropdown renders above main content Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add AI assistant with in-session copilot and standalone chat with RAG Implements three-phase AI assistant feature: - Phase 0: RAG infrastructure with pgvector embeddings, Voyage AI integration, tree chunking service, and semantic search over team's flow library - Phase 1: In-session copilot panel during flow navigation with contextual AI help, current step awareness, and suggested related flows - Phase 2: Standalone AI chat page with persistent conversation history, pin/delete, and configurable retention policies (account-level) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add account management, email verification, AI fixes, and user guides - Profile settings, account transfer, delete/leave account flows - Email verification with JWT tokens and Resend integration - AI assistant/copilot fixes: markdown rendering, shared RAG helpers, token tracking, input refocus, model_validate usage - User guides hub + detail pages with 13 topic guides - Sidebar and top bar navigation for guides Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: prevent stale chunk errors after deployments - Set Cache-Control no-cache on index.html in nginx so browsers always fetch fresh chunk references after a deploy - Auto-reload on chunk load failures (stale deploy detection) with loop prevention via sessionStorage - Show friendly "App Updated" message if auto-reload doesn't resolve it Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add email verification toggle to admin settings Adds platform-level toggle to enable/disable email verification. When disabled, the verification banner is hidden and the send endpoint returns 403. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-04 20:44:25 -05:00
parent 554ad84d9e
commit 4d2c4930fd
100 changed files with 8181 additions and 466 deletions
--- a/backend/app/services/assistant_chat_service.py
+++ b/backend/app/services/assistant_chat_service.py
@@ -0,0 +1,122 @@
+"""Standalone AI assistant chat service with RAG context.
+
+Provides persistent conversation history for general IT questions
+with semantic search over the team's flow library.
+"""
+import logging
+from typing import Optional, Any
+from uuid import UUID
+
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.core.ai_provider import get_ai_provider
+from app.models.assistant_chat import AssistantChat
+from app.services.rag_service import search as rag_search, build_rag_context, extract_suggested_flows
+
+logger = logging.getLogger(__name__)
+
+ASSISTANT_SYSTEM_PROMPT = """You are a Senior Systems and Network Engineer with 15+ years of experience working in Managed Service Provider (MSP) environments. You specialize in:
+- Windows Server, Active Directory, Group Policy, and Hybrid Identity (Entra ID)
+- Networking (TCP/IP, DNS, DHCP, VPN, firewall troubleshooting, Cisco/Fortinet)
+- Virtualization (VMware, Hyper-V) and cloud platforms (Azure, AWS, M365)
+- Endpoint management, RMM tools, and PSA platforms (ConnectWise, Datto, Kaseya)
+- PowerShell scripting and automation
+
+When answering:
+- Be direct and actionable — MSP engineers need fast, practical answers
+- Include specific commands, paths, and config values when relevant
+- Mention potential risks or gotchas before suggesting changes
+- If a relevant troubleshooting flow exists in the team's library, reference it
+- Keep responses concise but thorough — prefer bullet points and code blocks
+- Format code with proper markdown code blocks
+"""
+
+
+def _auto_title(message: str) -> str:
+    """Generate a short title from the first user message."""
+    title = message.strip()[:100]
+    if len(message) > 100:
+        title = title.rsplit(" ", 1)[0] + "..."
+    return title
+
+
+async def create_chat(
+    user_id: UUID,
+    account_id: UUID,
+    db: AsyncSession,
+) -> AssistantChat:
+    """Create a new empty chat."""
+    chat = AssistantChat(
+        user_id=user_id,
+        account_id=account_id,
+        messages=[],
+    )
+    db.add(chat)
+    await db.flush()
+    return chat
+
+
+async def send_message(
+    chat_id: UUID,
+    user_id: UUID,
+    account_id: UUID,
+    message: str,
+    db: AsyncSession,
+) -> tuple[str, list[dict[str, Any]], AssistantChat]:
+    """Send a user message and get AI response.
+
+    Returns (ai_content, suggested_flows, chat).
+    """
+    result = await db.execute(
+        select(AssistantChat).where(
+            AssistantChat.id == chat_id,
+            AssistantChat.user_id == user_id,
+        )
+    )
+    chat = result.scalar_one_or_none()
+    if not chat:
+        raise ValueError("Chat not found")
+
+    # Auto-title from first message
+    if chat.message_count == 0:
+        chat.title = _auto_title(message)
+
+    # RAG search
+    rag_results = await rag_search(
+        query=message,
+        account_id=account_id,
+        db=db,
+        limit=8,
+    )
+
+    # Build system prompt
+    system_prompt = ASSISTANT_SYSTEM_PROMPT + build_rag_context(rag_results)
+
+    # Build messages for AI
+    ai_messages = []
+    for msg in chat.messages:
+        if msg["role"] in ("user", "assistant"):
+            ai_messages.append({"role": msg["role"], "content": msg["content"]})
+    ai_messages.append({"role": "user", "content": message})
+
+    # Call AI
+    provider = get_ai_provider()
+    ai_content, input_tokens, output_tokens = await provider.generate_text(
+        system_prompt=system_prompt,
+        messages=ai_messages,
+        max_tokens=4096,
+    )
+
+    # Update chat
+    msgs = list(chat.messages)
+    msgs.append({"role": "user", "content": message})
+    msgs.append({"role": "assistant", "content": ai_content})
+    chat.messages = msgs
+    chat.message_count += 2
+    chat.total_input_tokens += input_tokens
+    chat.total_output_tokens += output_tokens
+
+    suggested_flows = extract_suggested_flows(rag_results)
+
+    return ai_content, suggested_flows, chat
--- a/backend/app/services/copilot_service.py
+++ b/backend/app/services/copilot_service.py
@@ -0,0 +1,202 @@
+"""Copilot service — in-session AI assistant with RAG context.
+
+Builds system prompts with current flow context and RAG results,
+manages conversation state, and returns AI responses with flow suggestions.
+"""
+import logging
+from datetime import datetime, timezone, timedelta
+from typing import Optional, Any
+from uuid import UUID
+
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession
+from sqlalchemy.orm import selectinload
+
+from app.core.ai_provider import get_ai_provider
+from app.models.tree import Tree
+from app.models.copilot_conversation import CopilotConversation
+from app.services.rag_service import search as rag_search, build_rag_context, extract_suggested_flows
+
+logger = logging.getLogger(__name__)
+
+COPILOT_SYSTEM_PROMPT = """You are a Senior Systems and Network Engineer with 15+ years of experience working in Managed Service Provider (MSP) environments. You specialize in:
+- Windows Server, Active Directory, Group Policy, and Hybrid Identity (Entra ID)
+- Networking (TCP/IP, DNS, DHCP, VPN, firewall troubleshooting, Cisco/Fortinet)
+- Virtualization (VMware, Hyper-V) and cloud platforms (Azure, AWS, M365)
+- Endpoint management, RMM tools, and PSA platforms (ConnectWise, Datto, Kaseya)
+- PowerShell scripting and automation
+
+You are acting as an in-session copilot while the user navigates a troubleshooting or procedural flow. You can see the flow context and their current position.
+
+When answering:
+- Be direct and actionable — MSP engineers need fast, practical answers
+- Include specific commands, paths, and config values when relevant
+- Mention potential risks or gotchas before suggesting changes
+- If a relevant troubleshooting flow exists in the team's library, reference it
+- Keep responses concise but thorough — prefer bullet points and code blocks
+"""
+
+
+def _build_flow_context(tree: Tree, current_node_id: Optional[str]) -> str:
+    """Build flow context string for the system prompt."""
+    parts = [
+        f"\n--- CURRENT FLOW CONTEXT ---",
+        f"Flow: {tree.name}",
+        f"Type: {tree.tree_type}",
+    ]
+    if tree.description:
+        parts.append(f"Description: {tree.description}")
+
+    if current_node_id and tree.tree_structure:
+        node = _find_node(tree.tree_structure, current_node_id)
+        if node:
+            parts.append(f"Current node type: {node.get('type', 'unknown')}")
+            parts.append(f"Current node: {node.get('content', node.get('label', 'Unknown'))}")
+            # Add options if it's a question/decision node
+            children = node.get("children", [])
+            if children and isinstance(children, list):
+                option_labels = [
+                    c.get("label", c.get("content", ""))
+                    for c in children if isinstance(c, dict)
+                ]
+                if option_labels:
+                    parts.append(f"Available options: {', '.join(option_labels)}")
+
+    return "\n".join(parts)
+
+
+def _find_node(structure: dict, node_id: str) -> Optional[dict]:
+    """Recursively find a node by ID in tree structure."""
+    if structure.get("id") == node_id:
+        return structure
+    for child in structure.get("children", []):
+        if isinstance(child, dict):
+            found = _find_node(child, node_id)
+            if found:
+                return found
+    # Check steps array for procedural flows
+    for step in structure.get("steps", []):
+        if isinstance(step, dict):
+            found = _find_node(step, node_id)
+            if found:
+                return found
+    return None
+
+
+async def start_conversation(
+    user_id: UUID,
+    account_id: UUID,
+    tree_id: UUID,
+    session_id: Optional[UUID],
+    current_node_id: Optional[str],
+    db: AsyncSession,
+) -> tuple[CopilotConversation, str]:
+    """Start a new copilot conversation.
+
+    Returns (conversation, greeting_message).
+    """
+    # Load tree
+    result = await db.execute(
+        select(Tree).options(selectinload(Tree.tags)).where(Tree.id == tree_id)
+    )
+    tree = result.scalar_one_or_none()
+    if not tree:
+        raise ValueError(f"Tree {tree_id} not found")
+
+    conversation = CopilotConversation(
+        user_id=user_id,
+        account_id=account_id,
+        tree_id=tree_id,
+        session_id=session_id,
+        current_node_id=current_node_id,
+        messages=[],
+        expires_at=datetime.now(timezone.utc) + timedelta(hours=24),
+    )
+    db.add(conversation)
+    await db.flush()
+
+    greeting = f"I'm your copilot for this **{tree.tree_type}** flow: **{tree.name}**. Ask me anything about the current step, alternative approaches, or related troubleshooting tips."
+
+    conversation.messages = [{"role": "assistant", "content": greeting}]
+    conversation.message_count = 1
+
+    return conversation, greeting
+
+
+async def send_message(
+    conversation_id: UUID,
+    user_id: UUID,
+    message: str,
+    current_node_id: Optional[str],
+    db: AsyncSession,
+) -> tuple[str, list[dict[str, Any]], CopilotConversation]:
+    """Send a user message and get AI response.
+
+    Returns (ai_content, suggested_flows, conversation).
+    """
+    result = await db.execute(
+        select(CopilotConversation).where(
+            CopilotConversation.id == conversation_id,
+            CopilotConversation.user_id == user_id,
+        )
+    )
+    conversation = result.scalar_one_or_none()
+    if not conversation:
+        raise ValueError("Conversation not found")
+
+    if conversation.expires_at < datetime.now(timezone.utc):
+        raise ValueError("Conversation has expired")
+
+    # Load tree for context
+    tree_result = await db.execute(
+        select(Tree).options(selectinload(Tree.tags)).where(Tree.id == conversation.tree_id)
+    )
+    tree = tree_result.scalar_one_or_none()
+    if not tree:
+        raise ValueError("Associated flow not found")
+
+    # Update current node
+    if current_node_id:
+        conversation.current_node_id = current_node_id
+
+    # RAG search
+    rag_results = await rag_search(
+        query=message,
+        account_id=conversation.account_id,
+        db=db,
+        limit=8,
+    )
+
+    # Build system prompt
+    system_prompt = COPILOT_SYSTEM_PROMPT
+    system_prompt += _build_flow_context(tree, conversation.current_node_id)
+    system_prompt += build_rag_context(rag_results)
+
+    # Build messages for AI
+    ai_messages = []
+    for msg in conversation.messages:
+        if msg["role"] in ("user", "assistant"):
+            ai_messages.append({"role": msg["role"], "content": msg["content"]})
+    ai_messages.append({"role": "user", "content": message})
+
+    # Call AI
+    provider = get_ai_provider()
+    ai_content, input_tokens, output_tokens = await provider.generate_text(
+        system_prompt=system_prompt,
+        messages=ai_messages,
+        max_tokens=2048,
+    )
+
+    # Update conversation
+    msgs = list(conversation.messages)
+    msgs.append({"role": "user", "content": message})
+    msgs.append({"role": "assistant", "content": ai_content})
+    conversation.messages = msgs
+    conversation.message_count += 2
+    conversation.total_input_tokens += input_tokens
+    conversation.total_output_tokens += output_tokens
+
+    # Extract suggested flows
+    suggested_flows = extract_suggested_flows(rag_results, exclude_tree_id=tree.id)
+
+    return ai_content, suggested_flows, conversation
--- a/backend/app/services/embedding_service.py
+++ b/backend/app/services/embedding_service.py
@@ -0,0 +1,78 @@
+"""Embedding provider abstraction for RAG.
+
+Uses Voyage AI (voyage-3.5, 1024 dims) as the embedding provider.
+Supports document and query input types for asymmetric search.
+"""
+import logging
+from typing import Optional
+
+from app.core.config import settings
+
+logger = logging.getLogger(__name__)
+
+
+async def get_embedding(
+    text: str,
+    input_type: str = "document",
+) -> Optional[list[float]]:
+    """Get embedding vector for text using Voyage AI.
+
+    Args:
+        text: The text to embed.
+        input_type: "document" for indexing, "query" for search queries.
+
+    Returns:
+        List of floats (1024 dims) or None if embedding service unavailable.
+    """
+    if not settings.VOYAGE_API_KEY:
+        logger.warning("VOYAGE_API_KEY not set — embedding service unavailable")
+        return None
+
+    try:
+        import voyageai
+
+        client = voyageai.AsyncClient(api_key=settings.VOYAGE_API_KEY)
+        result = await client.embed(
+            texts=[text],
+            model=settings.EMBEDDING_MODEL,
+            input_type=input_type,
+        )
+        return result.embeddings[0]
+    except Exception as e:
+        logger.error("Embedding failed: %s", e)
+        return None
+
+
+async def get_embeddings_batch(
+    texts: list[str],
+    input_type: str = "document",
+) -> Optional[list[list[float]]]:
+    """Get embedding vectors for multiple texts in a single API call.
+
+    Args:
+        texts: List of texts to embed.
+        input_type: "document" for indexing, "query" for search queries.
+
+    Returns:
+        List of embedding vectors or None if service unavailable.
+    """
+    if not texts:
+        return []
+
+    if not settings.VOYAGE_API_KEY:
+        logger.warning("VOYAGE_API_KEY not set — embedding service unavailable")
+        return None
+
+    try:
+        import voyageai
+
+        client = voyageai.AsyncClient(api_key=settings.VOYAGE_API_KEY)
+        result = await client.embed(
+            texts=texts,
+            model=settings.EMBEDDING_MODEL,
+            input_type=input_type,
+        )
+        return result.embeddings
+    except Exception as e:
+        logger.error("Batch embedding failed: %s", e)
+        return None
--- a/backend/app/services/rag_service.py
+++ b/backend/app/services/rag_service.py
@@ -0,0 +1,209 @@
+"""RAG service — index trees and search embeddings for AI context.
+
+Orchestrates tree chunking, embedding, and semantic search over the
+team's flow library via pgvector cosine similarity.
+"""
+import logging
+from typing import Optional, Any
+from uuid import UUID
+
+from sqlalchemy import text, delete
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.models.tree import Tree
+from app.models.tree_embedding import TreeEmbedding
+from app.services.embedding_service import get_embedding, get_embeddings_batch
+from app.services.tree_chunker import chunk_tree
+
+logger = logging.getLogger(__name__)
+
+
+async def index_tree(tree_id: UUID, db: AsyncSession) -> int:
+    """Chunk and embed a tree, storing results in tree_embeddings.
+
+    Deletes existing embeddings for this tree before re-indexing.
+    Returns the number of chunks indexed.
+    """
+    from sqlalchemy import select
+    from sqlalchemy.orm import selectinload
+
+    result = await db.execute(
+        select(Tree)
+        .options(selectinload(Tree.tags))
+        .where(Tree.id == tree_id)
+    )
+    tree = result.scalar_one_or_none()
+    if not tree:
+        logger.warning("index_tree: tree %s not found", tree_id)
+        return 0
+
+    # Delete existing embeddings
+    await db.execute(
+        delete(TreeEmbedding).where(TreeEmbedding.tree_id == tree_id)
+    )
+
+    # Chunk the tree
+    tag_names = [t.name for t in tree.tags] if tree.tags else []
+    chunks = chunk_tree(
+        tree_name=tree.name,
+        tree_type=tree.tree_type,
+        description=tree.description,
+        tags=tag_names,
+        tree_structure=tree.tree_structure,
+    )
+
+    if not chunks:
+        logger.info("index_tree: no chunks for tree %s", tree_id)
+        return 0
+
+    # Get embeddings for all chunks in batch
+    texts = [c["chunk_text"] for c in chunks]
+    embeddings = await get_embeddings_batch(texts, input_type="document")
+
+    if embeddings is None:
+        logger.warning("index_tree: embedding service unavailable for tree %s", tree_id)
+        return 0
+
+    # Insert embeddings
+    for chunk, embedding in zip(chunks, embeddings):
+        embedding_str = "[" + ",".join(str(v) for v in embedding) + "]"
+        await db.execute(
+            text("""
+                INSERT INTO tree_embeddings
+                    (tree_id, account_id, chunk_type, node_type, node_id, chunk_text, embedding_model, embedding, meta)
+                VALUES
+                    (:tree_id, :account_id, :chunk_type, :node_type, :node_id, :chunk_text, :embedding_model, :embedding::vector, :meta::jsonb)
+            """),
+            {
+                "tree_id": str(tree_id),
+                "account_id": str(tree.account_id) if tree.account_id else None,
+                "chunk_type": chunk["chunk_type"],
+                "node_type": chunk.get("node_type"),
+                "node_id": chunk.get("node_id"),
+                "chunk_text": chunk["chunk_text"],
+                "embedding_model": "voyage-3.5",
+                "embedding": embedding_str,
+                "meta": "{}",
+            },
+        )
+
+    logger.info("index_tree: indexed %d chunks for tree %s", len(chunks), tree_id)
+    return len(chunks)
+
+
+async def delete_tree_embeddings(tree_id: UUID, db: AsyncSession) -> None:
+    """Delete all embeddings for a tree."""
+    await db.execute(
+        delete(TreeEmbedding).where(TreeEmbedding.tree_id == tree_id)
+    )
+
+
+async def search(
+    query: str,
+    account_id: UUID,
+    db: AsyncSession,
+    limit: int = 8,
+    exclude_tree_id: Optional[UUID] = None,
+) -> list[dict[str, Any]]:
+    """Semantic search over team's flow library.
+
+    Args:
+        query: Natural language search query.
+        account_id: Scope search to team's flows.
+        db: Database session.
+        limit: Max results to return.
+        exclude_tree_id: Exclude chunks from this tree (for copilot context).
+
+    Returns:
+        List of dicts with tree_id, tree_name, tree_type, chunk_text, chunk_type, similarity.
+    """
+    query_embedding = await get_embedding(query, input_type="query")
+    if query_embedding is None:
+        return []
+
+    embedding_str = "[" + ",".join(str(v) for v in query_embedding) + "]"
+
+    exclude_clause = ""
+    params: dict[str, Any] = {
+        "embedding": embedding_str,
+        "account_id": str(account_id),
+        "limit": limit,
+    }
+
+    if exclude_tree_id:
+        exclude_clause = "AND te.tree_id != :exclude_tree_id"
+        params["exclude_tree_id"] = str(exclude_tree_id)
+
+    result = await db.execute(
+        text(f"""
+            SELECT
+                te.tree_id,
+                t.name as tree_name,
+                t.tree_type,
+                te.chunk_text,
+                te.chunk_type,
+                te.node_id,
+                1 - (te.embedding <=> :embedding::vector) as similarity
+            FROM tree_embeddings te
+            JOIN trees t ON t.id = te.tree_id
+            WHERE te.account_id = :account_id
+                AND t.deleted_at IS NULL
+                {exclude_clause}
+            ORDER BY te.embedding <=> :embedding::vector
+            LIMIT :limit
+        """),
+        params,
+    )
+
+    rows = result.mappings().all()
+    return [
+        {
+            "tree_id": str(row["tree_id"]),
+            "tree_name": row["tree_name"],
+            "tree_type": row["tree_type"],
+            "chunk_text": row["chunk_text"],
+            "chunk_type": row["chunk_type"],
+            "node_id": row["node_id"],
+            "similarity": float(row["similarity"]),
+        }
+        for row in rows
+    ]
+
+
+def build_rag_context(rag_results: list[dict[str, Any]]) -> str:
+    """Format RAG results into a system prompt section."""
+    if not rag_results:
+        return ""
+
+    parts = ["\n--- RELEVANT FLOWS FROM TEAM LIBRARY ---"]
+    for r in rag_results[:5]:  # Cap at 5 for prompt size
+        parts.append(f"- [{r['tree_type']}] {r['tree_name']}: {r['chunk_text'][:200]}")
+
+    return "\n".join(parts)
+
+
+def extract_suggested_flows(
+    rag_results: list[dict[str, Any]],
+    exclude_tree_id: Optional[UUID] = None,
+) -> list[dict[str, Any]]:
+    """Extract unique suggested flows from RAG results."""
+    seen_tree_ids: set[str] = set()
+    suggestions = []
+
+    for r in rag_results:
+        tid = r["tree_id"]
+        if exclude_tree_id and tid == str(exclude_tree_id):
+            continue
+        if tid in seen_tree_ids:
+            continue
+        if r["similarity"] < 0.3:
+            continue
+        seen_tree_ids.add(tid)
+        suggestions.append({
+            "tree_id": tid,
+            "tree_name": r["tree_name"],
+            "tree_type": r["tree_type"],
+            "relevance_snippet": r["chunk_text"][:150],
+        })
+
+    return suggestions[:3]
--- a/backend/app/services/retention_cleanup.py
+++ b/backend/app/services/retention_cleanup.py
@@ -0,0 +1,84 @@
+"""Chat retention cleanup job.
+
+Runs daily via APScheduler to enforce account-level retention settings:
+- Delete non-pinned chats older than chat_retention_days
+- Delete oldest non-pinned chats when count exceeds chat_retention_max_count
+"""
+import logging
+from datetime import datetime, timezone, timedelta
+
+from sqlalchemy import select, delete, func
+
+from app.core.database import async_session_maker
+from app.models.account import Account
+from app.models.assistant_chat import AssistantChat
+
+logger = logging.getLogger(__name__)
+
+
+async def cleanup_expired_chats() -> None:
+    """Enforce chat retention policies for all accounts."""
+    async with async_session_maker() as db:
+        try:
+            result = await db.execute(select(Account))
+            accounts = result.scalars().all()
+
+            total_deleted = 0
+            for account in accounts:
+                deleted = await _cleanup_account_chats(account, db)
+                total_deleted += deleted
+
+            await db.commit()
+            if total_deleted > 0:
+                logger.info("[retention] Cleaned up %d expired chats", total_deleted)
+        except Exception as e:
+            logger.error("[retention] Chat cleanup failed: %s", e)
+            await db.rollback()
+
+
+async def _cleanup_account_chats(account: Account, db) -> int:
+    """Enforce retention for a single account. Returns count deleted."""
+    deleted = 0
+
+    # Age-based retention
+    if account.chat_retention_days:
+        cutoff = datetime.now(timezone.utc) - timedelta(days=account.chat_retention_days)
+        result = await db.execute(
+            delete(AssistantChat)
+            .where(
+                AssistantChat.account_id == account.id,
+                AssistantChat.pinned == False,  # noqa: E712
+                AssistantChat.updated_at < cutoff,
+            )
+            .returning(AssistantChat.id)
+        )
+        deleted += len(result.all())
+
+    # Count-based retention
+    if account.chat_retention_max_count:
+        total = await db.scalar(
+            select(func.count(AssistantChat.id)).where(
+                AssistantChat.account_id == account.id,
+            )
+        ) or 0
+
+        if total > account.chat_retention_max_count:
+            excess = total - account.chat_retention_max_count
+            # Get oldest non-pinned chat IDs
+            oldest = await db.execute(
+                select(AssistantChat.id)
+                .where(
+                    AssistantChat.account_id == account.id,
+                    AssistantChat.pinned == False,  # noqa: E712
+                )
+                .order_by(AssistantChat.updated_at.asc())
+                .limit(excess)
+            )
+            ids_to_delete = [row[0] for row in oldest.all()]
+            if ids_to_delete:
+                await db.execute(
+                    delete(AssistantChat).where(AssistantChat.id.in_(ids_to_delete))
+                )
+                deleted += len(ids_to_delete)
+
+    return deleted
--- a/backend/app/services/tree_chunker.py
+++ b/backend/app/services/tree_chunker.py
@@ -0,0 +1,165 @@
+"""Tree chunker — converts tree_structure JSON into embeddable text chunks.
+
+Produces three chunk types:
+- tree_summary: Name + description + tags + type overview
+- node: Individual node content with breadcrumb path context
+- solution: Full solution/action text with path context
+"""
+import logging
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+def _get_breadcrumb(node: dict, parent_path: str = "") -> str:
+    """Build a breadcrumb path string for a node."""
+    content = node.get("content", node.get("label", ""))[:80]
+    if parent_path:
+        return f"{parent_path} > {content}"
+    return content
+
+
+def _chunk_node(
+    node: dict,
+    tree_name: str,
+    tree_type: str,
+    tags: list[str],
+    parent_path: str = "",
+) -> list[dict[str, Any]]:
+    """Recursively chunk a node and its children."""
+    chunks = []
+    node_type = node.get("type", "unknown")
+    node_id = node.get("id", "")
+    content = node.get("content", node.get("label", ""))
+    breadcrumb = _get_breadcrumb(node, parent_path)
+
+    # Build chunk text based on node type
+    if node_type in ("question", "decision"):
+        options = node.get("children", [])
+        option_labels = [
+            child.get("label", child.get("content", ""))[:100]
+            for child in options
+            if isinstance(child, dict)
+        ]
+        text_parts = [
+            f"[{node_type}] {content}",
+        ]
+        if option_labels:
+            text_parts.append(f"Options: {', '.join(option_labels)}")
+        text_parts.append(f"Path: {breadcrumb}")
+        text_parts.append(f"Flow: {tree_name} | Type: {tree_type}")
+        if tags:
+            text_parts.append(f"Tags: {', '.join(tags)}")
+
+        chunks.append({
+            "chunk_type": "node",
+            "node_type": node_type,
+            "node_id": node_id,
+            "chunk_text": "\n".join(text_parts),
+        })
+
+    elif node_type in ("action", "solution", "info", "warning"):
+        text_parts = [
+            f"[{node_type}] {content}",
+            f"Path: {breadcrumb}",
+            f"Flow: {tree_name} | Type: {tree_type}",
+        ]
+        if tags:
+            text_parts.append(f"Tags: {', '.join(tags)}")
+
+        chunk_type = "solution" if node_type == "solution" else "node"
+        chunks.append({
+            "chunk_type": chunk_type,
+            "node_type": node_type,
+            "node_id": node_id,
+            "chunk_text": "\n".join(text_parts),
+        })
+
+    elif node_type in ("step", "section_header"):
+        text_parts = [
+            f"[{node_type}] {content}",
+            f"Path: {breadcrumb}",
+            f"Flow: {tree_name} | Type: {tree_type}",
+        ]
+        if node.get("description"):
+            text_parts.insert(1, node["description"])
+        if tags:
+            text_parts.append(f"Tags: {', '.join(tags)}")
+
+        chunks.append({
+            "chunk_type": "node",
+            "node_type": node_type,
+            "node_id": node_id,
+            "chunk_text": "\n".join(text_parts),
+        })
+
+    # Recurse into children
+    children = node.get("children", [])
+    if isinstance(children, list):
+        for child in children:
+            if isinstance(child, dict):
+                chunks.extend(
+                    _chunk_node(child, tree_name, tree_type, tags, breadcrumb)
+                )
+
+    # Follow next_node_id linked nodes (action nodes)
+    # These are handled at the tree level, not recursively
+
+    return chunks
+
+
+def chunk_tree(
+    tree_name: str,
+    tree_type: str,
+    description: str | None,
+    tags: list[str],
+    tree_structure: dict[str, Any],
+) -> list[dict[str, Any]]:
+    """Convert a tree into embeddable text chunks.
+
+    Args:
+        tree_name: Name of the flow.
+        tree_type: troubleshooting | procedural | maintenance.
+        description: Flow description.
+        tags: List of tag names.
+        tree_structure: The tree_structure JSONB content.
+
+    Returns:
+        List of chunk dicts with keys: chunk_type, node_type, node_id, chunk_text.
+    """
+    chunks = []
+
+    # Tree summary chunk
+    summary_parts = [
+        f"Flow: {tree_name}",
+        f"Type: {tree_type}",
+    ]
+    if description:
+        summary_parts.append(f"Description: {description}")
+    if tags:
+        summary_parts.append(f"Tags: {', '.join(tags)}")
+
+    chunks.append({
+        "chunk_type": "tree_summary",
+        "node_type": None,
+        "node_id": None,
+        "chunk_text": "\n".join(summary_parts),
+    })
+
+    # Chunk the tree structure nodes
+    root = tree_structure
+    if isinstance(root, dict):
+        # Handle both flat structure and nested
+        if "children" in root or "type" in root:
+            chunks.extend(
+                _chunk_node(root, tree_name, tree_type, tags)
+            )
+        # Handle steps array (procedural flows)
+        if "steps" in root and isinstance(root["steps"], list):
+            for step in root["steps"]:
+                if isinstance(step, dict):
+                    chunks.extend(
+                        _chunk_node(step, tree_name, tree_type, tags)
+                    )
+
+    return chunks