adding in entity extraction layer

packages/memory-service/src/entities/extraction.js

@@ -0,0 +1,67 @@
+const { getEnv } = require('@nexusai/shared');
+const { upsertEntity } = require('./index');
+
+const EXTRACTION_URL = getEnv('EXTRACTION_URL', 'http://localhost:11434');
+const EXTRACTION_MODEL = getEnv('EXTRACTION_MODEL', 'qwen2.5:3b');
+
+const ENTITY_TYPES = ['person', 'place', 'project', 'technology', 'concept', 'organization'];
+
+function buildExtractionPrompt(userMessage, aiResponse) {
+    return [
+        'Extract named entities from the conversation below.',
+        `Valid entity types: ${ENTITY_TYPES.join(', ')}.`,
+        'Return ONLY a JSON array. No explanation, no markdown, no code fences.',
+        'Each item must have exactly these fields: "name" (string), "type" (one of the valid types), "notes" (one sentence fact about this entity from the conversation).',
+        'If no entities are found, return: []',
+        '',
+        `User: ${userMessage}`,
+        `Assistant: ${aiResponse}`,
+        '',
+        'JSON array:',
+    ].join('\n');
+}
+
+async function extractAndStoreEntities(userMessage, aiResponse) {
+    try {
+        const res = await fetch(`${EXTRACTION_URL}/api/generate`, {
+            method: 'POST',
+            headers: { 'Content-Type': 'application/json' },
+            body: JSON.stringify({
+                model: EXTRACTION_MODEL,
+                prompt: buildExtractionPrompt(userMessage, aiResponse),
+                stream: false,      // we want the complete response, not a stream
+                options: {
+                    temperature: 0.1,   // low temp for deterministic structured output
+                    num_predict: 512,
+                },
+            }),
+        });
+
+        if (!res.ok) throw new Error(`Ollama responded ${res.status}`);
+
+        const data = await res.json();
+        const raw = data.response?.trim() ?? '';
+
+        // Strip markdown fences defensively — small models sometimes add them anyway
+        const clean = raw.replace(/^```(?:json)?\n?/, '').replace(/\n?```$/, '').trim();
+        const entities = JSON.parse(clean);
+
+        if (!Array.isArray(entities)) throw new Error('Response was not a JSON array');
+
+        let saved = 0;
+        for (const { name, type, notes } of entities) {
+            // Skip anything malformed or with an unrecognised type
+            if (!name || !type || !ENTITY_TYPES.includes(type)) continue;
+            upsertEntity(name, type, notes ?? null);
+            saved++;
+        }
+
+        if (saved > 0) console.log(`[entities] Extracted and stored ${saved} entities`);
+
+    } catch (err) {
+        // Non-critical — log and move on, episode is already saved
+        console.warn('[entities] Extraction failed:', err.message);
+    }
+}
+
+module.exports = { extractAndStoreEntities };

packages/memory-service/src/episodic/index.js

@@ -1,6 +1,7 @@
 const {getDB} = require('../db');
 const { EPISODIC, getEnv, SERVICES, parseRow, formatEpisodeText } = require('@nexusai/shared');
 const semantic = require('../semantic');
+const { extractAndStoreEntities } = require('../entities/extraction')
 
 // --Sessions --------------------------------------------------
 
@@ -127,6 +128,10 @@ async function createEpisode(sessionId, userMessage, aiResponse, tokenCount = nu
     }))
     .catch(err => console.error(`Failed to embed episode ${episode.id}:`, err.message));
 
+  extractAndStoreEntities(userMessage, aiResponse)
+    .catch(err => console.error(`Failed to extract entities for episode ${episode.id}:`, err.message));
+
+
   return episode;
 }