Added semantic episode searching

packages/orchestration-service/src/chat/index.js

@@ -1,15 +1,28 @@
 const memory = require('../services/memory');
 const inference = require('../services/inference');
+const embedding = require('../services/embedding');
+const qdrant = require('../services/qdrant');
 
 const SYSTEM_PROMPT = `You are a helpful, context-aware AI assistant. 
 You have access to memories of past conversations with the user. 
 Use them to provide consistent, personalised responses.`;
 
 const RECENT_EPISODE_LIMIT = 10; // Number of recent episodes to retrieve for context
+const SEMANTIC_LIMIT = 5;
+const SCORE_THRESHOLD = 0.75;
 
-function buildPrompt(recentEpisodes, userMessage) {
+function buildPrompt(recentEpisodes, semanticEpisodes, userMessage) {
     const parts = [SYSTEM_PROMPT];
 
+    if (semanticEpisodes.length > 0 )
+    {
+        parts.push('Here are some relevant memories from earlier conversations:')
+        for (const ep of semanticEpisodes) {
+            parts.push(`User: ${ep.user_message}\nAssistant: ${ep.ai_response}`);
+        }
+        parts.push('---')
+    }
+
     if (recentEpisodes.length > 0) {
         parts.push(`Here are some relevant memories from your past conversations:`);
         for (const ep of recentEpisodes) {
@@ -24,26 +37,46 @@ function buildPrompt(recentEpisodes, userMessage) {
     return parts.join('\n');
 }
 
+async function getSemanticEpisodes(userMessage, sessionId, recentIds) {
+    try {
+        const vector = await embedding.embed(userMessage);
+        const results = await qdrant.searchEpisodes( vector, {
+            limit: SEMANTIC_LIMIT,
+            scoreThreshold: SCORE_THRESHOLD,
+            sessionId,
+        });
+
+        const fetched = await Promise.all(
+            results
+                .filter(r => !recentIds.has(r.id))
+                .map(r => memory.getEpisodeById(r.id))
+        );
+        return fetched.filter(Boolean);
+    } catch (err) {
+        console.warn(`[orchestration] Semantic search failed, continuing without: `, err.message);
+        return [];
+    }
+}
+
 async function chat(externalId, userMessage, options = {}) {
     // 1. Resolve or create session
     let session = await memory.getSessionByExternalId(externalId);
-    if (!session) {
+    if (!session) session = await memory.createSession(externalId);
-        session = await memory.createSession(externalId);
-    }
 
     // 2. Fetch recent episodes for context
-    const recentEpisodes = await memory.getRecentEpisodes(
+    const recentEpisodes = await memory.getRecentEpisodes(session.id, RECENT_EPISODE_LIMIT );
-        session.id,
+    const recentIds = new Set(recentEpisodes.map(e => e.id));
-        RECENT_EPISODE_LIMIT
-    );
 
-    // 3. Assemble prompt
+    // 3. Semantic Search
-    const prompt = buildPrompt(recentEpisodes, userMessage);
+    const semanticEpisodes = await getSemanticEpisodes(userMessage, session.id, recentIds);
 
-    // 4. Run inference
+    // 4. Assemble prompt
+    const prompt = buildPrompt(recentEpisodes, semanticEpisodes, userMessage);
+
+    // 5. Run inference
     const result = await inference.complete(prompt, options);
 
-    // 5. Write episode back to memory
+    // 6. Write episode back to memory
     memory.createEpisode(
         session.id,
         userMessage,
@@ -51,7 +84,7 @@ async function chat(externalId, userMessage, options = {}) {
         (result.evalCount || 0) + (result.promptEvalCount || 0 )
     ).catch(err => console.error(`[orchestration] Failed to save episode`, err.message));
 
-    // 6. Return response
+    // 7. Return response
     return {
         sessionId: externalId,
         response: result.text,

packages/orchestration-service/src/services/embedding.js

@@ -0,0 +1,18 @@
+const {getEnv, SERVICES } = require('@nexusai/shared')
+
+const BASE_URL = getEnv('EMBEDDING_SERVICE_URL', SERVICES.EMBEDDING_URL);
+
+async function embed(text) {
+    const res = await fetch(`${BASE_URL}/embed`, {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json'},
+        body: JSON.stringify({text}),
+    })
+
+    if (!res.ok) throw new Error(`Embedding service error: ${res.status}`);
+
+    const data = await res.json();
+    return data.embedding;
+}
+
+module.exports = { embed };

packages/orchestration-service/src/services/memory.js

@@ -40,9 +40,17 @@ async function createEpisode(sessionId, userMessage, aiResponse, tokenCount) {
     return res.json();
 }
 
+async function getEpisodeById(episodeId) {
+    const res = await (`${BASE_URL}/episodes/${episodeId}`);
+    if (res.status === 404) return null;
+    if (!res.ok) throw new Error(`Failed to fetch episode: ${res.status}`);
+    return res.json();
+}
+
 module.exports = {
     getSessionByExternalId,
     createSession,
     getRecentEpisodes,
-    createEpisode
+    createEpisode,
+    getEpisodeById
 }

packages/orchestration-service/src/services/qdrant.js

@@ -0,0 +1,27 @@
+const {getEnv, QDRANT, COLLECTIONS } = require('@nexusai/shared')
+
+const BASE_URL = getEnv('QDrant_URL', QDRANT.DEFAULT_URL);
+
+async function searchEpisodes( vector, {limit = 5, scoreThreshold = 0.75, sessionId } = {}) {
+    const body = {vector, limit, score_threshold: scoreThreshold, with_payload: true};
+
+    if (sessionId) {
+        body.filter = { must: [{key: 'sessionId', match: {value: sessionId} }] };
+    }
+
+    const res = await fetch (
+        `${BASE_URL}/collections/${COLLECTIONS.EPISODES}/points/search`,
+        {
+            method: 'POST',
+            headers: {'Content-Type': 'application/json'}, 
+            body: JSON.stringify(body)
+        }
+    );
+
+    if (!res.ok) throw new Error(`QDrant error: ${res.status}`);
+
+    const data = await res.json();
+    return data.result;
+}
+
+module.exports = { searchEpisodes };