Give your WhatsApp AI agent context with contact notes

Your customer asks about their order. Your AI agent should already know they're a VIP, that they ordered recently, or that they're waiting on a refund. You don't want to repeat yourself, and you don't want the agent to sound like it's meeting them for the first time.

In the new Whatsable console, your AI agent reads contact notes each time it replies. The context you type in the inbox and the context your automation writes both give the agent background, so it knows who it's talking to.

Two notes, one view for your agent

Every contact in Whatsable has two note fields. The team note is the one you type in the inbox when you want to leave a reminder for yourself or a colleague. The automation note is the field your Make, Zapier, or n8n flow writes to when it syncs data from your CRM, order system, or support desk.

Your AI agent reads both. Before it replies, it sees the team note and the automation note as background. You can give it the context it needs without building a complicated prompt or waiting for a human to jump in.

Each note holds up to 5,000 characters. The agent treats both notes the same way: as context about the person, not as instructions.

What the agent does with notes

Notes give the agent context. It uses them to understand who it's talking to, what's happened before, and what might matter in this conversation. Notes are never a source for prices or terms.

If your automation note says "VIP customer, ordered three times this month, prefers email for receipts," the agent knows that. If your team note says "waiting on refund, escalated to finance," the agent knows that too.

The agent reads only the two contact-level notes: the one you type and the one your automation writes.

How to use notes with your automation

The automation note is where your Make, Zapier, or n8n flow writes context from other systems. When a customer places an order, your flow can write order details to the automation note. When a support ticket closes, your flow can write the resolution.

The agent reads that note on every turn, so it knows what's happening even if the customer doesn't mention it. You write the context to the note field, and the agent sees it.

The automation note is never an instruction. If you write "tell the customer their order ships tomorrow," the agent won't do that. It will know the order ships tomorrow, and it will use that context if the customer asks. The note is background, not a command.

How to use the team note in the inbox

The team note is the one you type. It's the place to leave a reminder about a conversation, a customer preference, or something that matters for next time. If you know a customer always asks for tracking numbers, write that. If you know they're frustrated and need a careful reply, write that too.

Your AI agent reads the team note each time it replies, so the context you leave gives it background for the conversation. You don't need to repeat yourself, and you don't need to brief the agent every time the customer writes back.

Nothing to set up. Type the note, and the agent sees it.

Guardrails: what the agent won't do

Notes are context, not sources. The agent never quotes prices or terms that appear only in a note. If your automation writes that a customer paid for an annual plan, the agent knows that, but it won't quote the amount unless your knowledge base or prompt confirms it.

The agent won't treat a note as an instruction. If you write "offer a discount," the agent won't do that unless your prompt or knowledge base tells it to. The note is background, not a command.

FAQ

Does the agent read the thread of internal notes inside a conversation?

No. The agent reads the two contact-level notes: the team note you type in the inbox and the automation note your Make, Zapier, or n8n flow writes.

Can I use notes to change how the agent behaves?

No. Notes give the agent context. It reads them each time it replies to understand who it's talking to, but they are background, not instructions. Your prompt and knowledge base are where you configure the agent.

What happens if I write a price in a note?

The agent sees it as context, but it won't quote it. Notes are never a source for prices or terms. If your knowledge base or prompt confirms the price, the agent can say it. Otherwise, it stays in the background.