Enhance Realtime Prompting Guide with clarifications and new content

- Revised text for improved clarity, including updates to examples and instructions on language control and repetition.
- Added a new section on prompt optimization to help users refine their prompts for better model performance.
- Enhanced descriptions to emphasize the importance of clear instructions and the impact of language constraints on model behavior.

Author: minh-hoque

examples/Realtime_prompting_guide.ipynb

@@ -73,12 +73,11 @@
    "source": [
     "# General Tips\n",
     "- **Iterate relentlessly**: Small wording changes can make or break behavior.\n",
-    "  - Example: swapping “inaudible” → “intelligible” boosted noisy input handling.\n",
-    "- **Make instructions clear**: The realtime model is excellent at instruction following, so unclear or conflicting instructions will degrade performance.\n",
+    "  - Example: swapping “inaudible” → “intelligible” boosted noisy audio input handling.\n",
     "- **Prefer bullets over paragraphs**: Clear, short bullets outperform long paragraphs.\n",
     "- **Guide with examples**: The model strongly follows onto sample phrases.\n",
     "- **Be precise**: Ambiguity or conflicting instructions = degraded performance similar to GPT-5.\n",
-    "- **Control language**: Pin output to a target language if you see drift.\n",
+    "- **Control language**: Pin output to a target language if you see unwanted language switching.\n",
     "- **Fight repetition**: Add a Variety rule to reduce robotic phrasing.\n"
    ]
   },
@@ -520,6 +519,7 @@
    "metadata": {},
    "source": [
     "## Language Constraint\n",
+    "Language constraints ensure the model consistently responds in the intended language, even in challenging conditions like background noise or multilingual inputs.\n",
     "\n",
     "- **When to use**: To prevent accidental language switching in multilingual or noisy environments.\n",
     "- **What it does**: Locks output to the chosen language to prevent accidental language changes.\n",
@@ -617,7 +617,7 @@
    "metadata": {},
    "source": [
     "## Reduce Repetition\n",
-    "Prevents “robotic” outputs by encouraging varied phrasing while preserving meaning and brand voice.\n",
+    "The realtime model can follow sample phrases closely to stay on-brand, but it may overuse them, making responses sound robotic or repetitive. Adding a repetition rule helps maintain variety while preserving clarity and brand voice.\n",
     "\n",
     "- **When to use**: Outputs recycle the same openings, fillers, or sentence patterns across turns or sessions.\n",
     "- **What it does**: Adds a variety constraint—discourages repeated phrases, nudges synonyms and alternate sentence structures, and keeps required terms intact.\n",
@@ -657,7 +657,7 @@
    "id": "7d0da635",
    "metadata": {},
    "source": [
-    "This is the responses **before** applying the instruction using `gpt-realtime`\n",
+    "This is the responses **before** applying the instruction using `gpt-realtime`. The model repeats the same confirmation `Got it`.\n",
     "\n",
     "<img\n",
     "  src=\"../images/repeat_before.png\"\n",
@@ -818,7 +818,7 @@
    "id": "f5bc3747",
    "metadata": {},
    "source": [
-    "If you are following a conversation flow prompting strategy, you can specify which conversation state needs to apply the alpha-numeric pronunciations instruction.\n",
+    "*Tip: If you are following a conversation flow prompting strategy, you can specify which conversation state needs to apply the alpha-numeric pronunciations instruction.*\n",
     "\n",
     "### Example \n",
     "*(taken from the conversation flow of the prompt of our [openai-realtime-agents](https://github.com/openai/openai-realtime-agents/blob/main/src/app/agentConfigs/customerServiceRetail/authentication.ts))*\n",
@@ -882,7 +882,7 @@
    "metadata": {},
    "source": [
     "## Instruction Following\n",
-    "Like 4.1, if the instructions are conflicting, ambiguous or not clear, Realtime models will perform worse\n",
+    "Like GPT-4.1 and GPT-5, if the instructions are conflicting, ambiguous or not clear, the new realtime model will perform worse\n",
     "\n",
     "- **When to use**: Outputs drift from rules, skip phases, or misuse tools.\n",
     "- **What it does**: Uses an LLM to point out ambiguity, conflicts, and missing definitions before you ship.\n"
@@ -895,7 +895,7 @@
    "source": [
     "### **Instructions Quality Prompt (can be used in ChatGPT or with API)**\n",
     "\n",
-    "Use the following prompt to identify problematic areas in your prompt that you can fix.\n",
+    "Use the following prompt with GPT-5 to identify problematic areas in your prompt that you can fix.\n",
     "\n",
     "```\n",
     "## Role & Objective  \n",
@@ -930,6 +930,29 @@
     "```"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "e823fcd9",
+   "metadata": {},
+   "source": [
+    "### **Prompt Optimization Meta Prompt (can be used in ChatGPT or with API)**\n",
+    "\n",
+    "This meta-prompt helps you improve your base system prompt by targeting a specific failure mode. Provide the current prompt and describe the issue you’re seeing, the model (GPT-5) will suggest refined variants that tighten constraints and reduce the problem.\n",
+    "\n",
+    "```\n",
+    "Here's my current prompt to an LLM:\n",
+    "[BEGIN OF CURRENT PROMPT]\n",
+    "{CURRENT_PROMPT}\n",
+    "[END OF CURRENT PROMPT]\n",
+    " \n",
+    "But I see this issue happening from the LLM:\n",
+    "[BEGIN OF ISSUE]\n",
+    "{ISSUE}\n",
+    "[END OF ISSUE]\n",
+    "Can you provide some variants of the prompt so that the model can better understand the constraints to alleviate the issue?\n",
+    "```"
+   ]
+  },
   {
    "cell_type": "markdown",
    "id": "e9d05945",