Optional Power-Ups — Extra Nodes

Your prompt map is already ready with the 5 core nodes. These are optional upgrades.

3 min read

Your map is already good to go

If you've filled in the five core nodes — Identity, Style, Knowledge, Rules, and Examples — your agent is ready to use. Everything from here is optional. Think of these as nice-to-have upgrades, not requirements.

You can always come back and add these later. There's no rush, and your agent will work perfectly fine without them.

How to add an extra node

1
Click the "+" button on the left edge of your canvas.
2
Drag any node from the menu onto the canvas.
3
It connects automatically to your prompt map. No wiring needed.

Tools Node

If your agent has access to external tools — like looking up an order status, searching a database, or browsing the web — you list them here. The Tools node tells the agent what it can reach for beyond just the conversation.

Most people don't need this right away. It's mainly for setups where your agent is connected to other software. If you're not sure whether you need it, you probably don't yet. You'll know when you do.

Memory Node

This controls whether your agent remembers things. There are two levels: should it remember what was said earlier in the current conversation? (Usually yes — otherwise it forgets what the customer just told it two messages ago.) And should it remember past conversations with the same person? That depends on your setup and your platform.

For most use cases, keeping current-conversation memory on is the right call. Past-conversation memory is more of a power-user feature that depends on how your agent is deployed.

Opening Message Node

This is the very first thing your agent says when someone starts a conversation — before the customer has typed anything. It's a great place to set expectations and offer quick options.

You can set up a quick-reply menu so customers see buttons like "Track my order," "Browse products," or "Talk to a human" right away. This helps people get to the right answer faster and reduces the "I don't know what to ask" problem. A good opening message feels like walking into a store where someone says "Hey, what can I help you with today?"

A simple opening message like "Hi! I can help with orders, product questions, or connect you with our team. What do you need?" goes a long way. Don't overthink it.

Closing Behavior Node

This controls how your agent wraps up a conversation. Should it ask for feedback? Summarize what was discussed? Just say goodbye? You decide.

Some businesses like to end with a satisfaction check ("Did that answer your question?"), while others prefer a clean sign-off. There's no wrong answer — it depends on your brand and what feels natural for your customers.

Context Variables Node (Advanced)

This one is for technical setups where your agent automatically receives information about the user — like their name, account type, or subscription plan — before the conversation even starts. It's useful when your agent is embedded in an app and your developer passes user data to it.

Skip this unless you have a developer involved in your setup. If you don't know what context variables are, you don't need this node right now.

Don't feel pressured to add any of these. Your five core nodes are doing the heavy lifting. These extras are here when you're ready for them — whether that's today, next week, or never.

Clicking the '+' button → dragging an Opening Message node onto the canvas → it auto-connecting → typing a quick greeting and setting up two quick-reply buttons

Video • 30 seconds

You just learned about the optional nodes you can add to your prompt map whenever you're ready. None of them are required — your agent works great with just the five core nodes. But when you want to level up, these power-ups are here waiting for you.