Skip to main content

What is Era?

Era turns a conversation into a short context block for your AI agent. The block summarizes the user’s current goal, intent, urgency, and recommended next step. Call turnContext() before each model response, then add the returned string to your system prompt:
The rest of your model call stays the same.

What Era returns

turnContext() returns a plain string:
Your model uses that block as internal context. Do not show it to the user.

How one turn works

  1. Your user sends a message.
  2. Your backend calls turnContext() with the conversation and visitor ID.
  3. Era returns a context block.
  4. You add the block to the system prompt and call your model.
  5. Your model responds with a better read on what the user needs.
In the default async mode, Era returns the previous turn’s prepared block. Use sync mode when you need context for the current turn and can accept more latency. Compare async and sync mode.

What Era is not

Era is not a replacement for conversation history or long-term memory.
  • Conversation history gives the model the exact messages.
  • Memory stores durable facts, such as a user’s preferences.
  • Era summarizes what the user is trying to do right now.
You can use all three together.

Which SDK do I need?

The secret Node SDK API key must stay on your backend. The Web and Flutter SDKs use publishable keys.

Next steps

Node SDK quickstart

Make your first Node SDK call.

Setup SDKs

Choose the SDKs for your application.

All SDK methods

Find every public method and function.

See an example

Compare the same conversation with and without Era.