diff --git a/index.html b/index.html
index 9a90f98..374bac0 100644
--- a/index.html
+++ b/index.html
@@ -13,26 +13,70 @@
   <body>
 
 <script>
+// used for keywords like LAST / PREVIOUS / ALL
+let addedContent = []
+
+function getAllPrimitives(){
+	const other_primitives = ["camera", "cursor", "sky", "light", "sound", "videosphere"]
+	const other_primitives_with_param_needed = ["text", "gltf-model", "obj-model", "troika-text"]
+	return Object.getOwnPropertyNames(AFRAME.primitives.primitives)
+	// thanks to https://github.com/Utopiah/aframe-inVR-blocks-based-editor/blob/master/aframe-invr-inspect.js
+		.map( i => i.replace("a-",""))
+		.filter( i => other_primitives.indexOf(i) < 0 )
+		.filter( i => other_primitives_with_param_needed.indexOf(i) < 0 ) // temporarilty disabled
+		// .map( (i,j) => addPrimitive( i, ""+ j/7 + " 1.4 -0.5" ) )
+		.map( (i,j) => i )
+} // adapted from https://git.benetou.fr/utopiah/text-code-xr-engine/src/commit/0e1f297ec0cd17b0356811dfa0ab55f1e2629e7c/index.html#L2101
+
 const SpeechRecognition = window.webkitSpeechRecognition; 
 recognizer = new SpeechRecognition();
 recognizer.interimResults = true;
 recognizer.continuous = true;
 recognizer.lang = 'en-US';
 
+const aframeprimitives = getAllPrimitives()
+const speechactions = [ "add", "apply" ]
+const speechcustomcomponents = [ "target", "teleporter" ]
+
 recognizer.onresult = (event) => {
    let result = event.results[event.resultIndex]
    if (result.isFinal) {
-	console.log('You said: ' + result[0].transcript + '_END')
+	console.log('You said: ' + result[0].transcript ) 
+
+	// should instead look if starting with speechContent following by aframeprimitives, optionally speechcustomcomponents
+		// then apply accordingly
+	let speechContent = result[0].transcript
+	let cmd_words = speechContent.split(" ").map( i => i.toLowerCase() )
+	if (cmd_words[0] == speechactions[0]){ // limited to add command for now
+		console.log("recognized proper command")
+		let primitive = cmd_words[1] // assuming fixed order for now
+		let el = document.createElement("a-"+primitive)
+		el.setAttribute("target", "")
+		el.setAttribute("scale", ".1 .1 .1")
+		el.setAttribute("position", "0 1.5 -0.4")
+		addedContent.push(el)
+		AFRAME.scenes[0].appendChild( el )
+	}
+
+	if (cmd_words[0] == speechactions[1]){ // limited to apply
+		console.log("recognized apply command")
+		addedContent.at(-1).setAttribute( cmd_words[1], cmd_words[2]) // assuming fixed order for now
+	}
+
 	if ( result[0].transcript == "Add red cube"){
-		// should instead look for syntax e.g ACTION PROPERTY_SHORTCUT ENTITY_TYPE
-			// itself generated from AFrame introspection
 		console.log("recognized command")
+		let newPostIt = addNewNoteAsPostItNote(speechContent, "0 1.2 -.5")
+			// could become jxr code proper later, also allowing to re-execute a command again
 		let el = document.createElement("a-box")
 		el.setAttribute("color", "red")
 		el.setAttribute("target", "")
 		el.setAttribute("scale", ".1 .1 .1")
 		el.setAttribute("position", "0 1.5 -0.4")
 		AFRAME.scenes[0].appendChild( el )
+
+		// among ACTION could also have a selection area, e.g a bounding box wireframe to apply the action to all elements within in
+		addedContent.push(el)
+
 	}
    }
 }
