---
title: "Aster by AsterWise Is Now on FlowRunner: One API Key, Two Models"
description: "Aster is an LLM router with two models, Aster Work and Aster Code. Pick one and Aster picks the underlying model for each request. Here is what it does in FlowRunner, a customer reply flow you can build in a few steps, and how to set it up."
url: https://flowrunner.ai/blog/aster-llm-router-on-flowrunner
date_published: 2026-11-02
date_modified: 2026-11-02T13:31:48-08:00
---

Product November 2, 2026 5 min read

# Aster by AsterWise Is Now on FlowRunner: One API Key, Two Models

Aster is an LLM router with two models, Aster Work and Aster Code. Pick one and Aster picks the underlying model for each request. Here is what it does in FlowRunner, a customer reply flow you can build in a few steps, and how to set it up.

![A FlowRunner New Integration card on an ochre field with the statement Introducing Aster integration. and the line One key. Two models. Now on FlowRunner.](https://flowrunner.ai/images/blog/aster-llm-router-on-flowrunner-hero.webp)

Every time you add AI to an automation, you have to pick a provider and a model. Then a better model comes out and you get to pick again, in every flow that uses the old one. [Aster by AsterWise](https://asterwise.dev) takes that choice off your plate, and as of today it’s on FlowRunner.

## What Aster is

Aster is an LLM router. It gives you one OpenAI-compatible API and two models to choose from:

-   **Aster Work** for writing, research and everyday tasks
-   **Aster Code** for coding and repository work

You tell Aster what kind of work it is, and Aster picks the underlying model for each request. You keep one API key and one balance, and when the models behind Aster change, your flows don’t have to.

## What you can do with it in FlowRunner

The [Aster integration](https://flowrunner.ai/integrations/aster) has two actions.

**Generate Text** is the one you’ll use most. You give it a prompt and, if you like, system instructions (the tone, the rules, what to leave out). It returns the answer as plain text, along with the finish reason and the token usage. Drafting a reply, summarizing a thread and explaining an error all fit here.

**Create Chat Completion** is for when you need more control. It takes a full conversation history, function tools the model can ask to call, a tool choice, reasoning effort, an output token limit and sampling controls. It returns the text, any tool calls the model wants to make, the finish reason and the usage. Reasoning effort follows the model: Aster Work accepts minimal through xhigh, and Aster Code accepts low or high.

A few things are taken care of for you. If you pick a setting the chosen model doesn’t support, the step refuses it before anything is sent, so you aren’t billed for a request that was going to fail. Error messages say what to fix, whether that’s a wrong key, a low balance or a daily spending limit. The timeout is adjustable from 1 to 600 seconds and defaults to 120. If a step times out, or you can’t tell whether a request went through, check Usage in the Aster app before you retry, because another request may be charged again.

The integration doesn’t stream, so each step gets one response rather than streamed chunks. That doesn’t mean the answer is always complete. The output token limit can cut it short, so check the finish reason: `length` means the answer was truncated.

## Walkthrough: a customer email, drafted by Aster and approved by a person

This is the flow we’d start with. A customer writes in, Aster Work drafts a reply, and someone on your team approves it before it goes out. Aster does the slow part, and a person still decides what your company says.

**1\. A new email arrives.** The flow starts on the [Gmail](https://flowrunner.ai/integrations/gmail-service) “On New Email” trigger, watching your support inbox. A form submission works just as well if enquiries come in through your website.

**2\. Aster drafts the reply.** Add the Aster “Generate Text” action and set the model to Aster Work. In the system instructions, write down how your team answers: the tone, what you never promise, how you sign off. For the prompt, pass in the sender, the subject and the body of the email. Aster returns the draft as plain text.

It’s worth checking the finish reason here. If it says the answer hit the token limit, the draft was cut off, and the reviewer should know that before reading it.

**3\. A person reviews it.** The flow posts the customer’s email and Aster’s draft to a [Slack](https://flowrunner.ai/integrations/slack) channel with Approve and Reject buttons, then waits. When someone clicks a button, the Slack “On Block Action” trigger picks the flow back up with their answer.

**4\. The reply goes out, or it doesn’t.** On Approve, Gmail “Send Message” sends the draft to the customer. On Reject, nothing is sent and the reviewer answers the customer themselves.

That last step is the reason this flow works for customer email. The AI can write the first draft of every reply, and nothing reaches a customer that a person hasn’t read. We call this [human-in-the-loop](https://flowrunner.ai/concepts/human-in-the-loop), and in FlowRunner it’s just a few more steps in the flow.

## More ideas

-   When a [GitHub](https://flowrunner.ai/integrations/github) issue is opened, Aster Code proposes a fix and the flow posts it as a comment on the issue.
-   When a support thread gets long, Aster summarizes it and the flow posts the summary to a Slack channel.
-   A tool-calling agent built on Create Chat Completion looks up the customer or the order in your CRM before it answers.

## Getting started

1.  **Create your Aster key.** Sign in at [app.asterwise.dev](https://app.asterwise.dev), create an API key and add a balance. Aster usage is billed to that balance.
2.  **Connect Aster in FlowRunner.** Add an Aster step to a flow and paste the key into its Aster connection.
3.  **Add the action.** Pick Generate Text or Create Chat Completion, choose Aster Work or Aster Code, and write your prompt.

AsterWise has its own [setup guide for FlowRunner](https://docs.asterwise.dev/integrations/flowrunner), listed in its docs alongside Claude Code, Codex and Cline.

## Try it

The [Aster integration page](https://flowrunner.ai/integrations/aster) shows both actions and more ways to use them. If you don’t have a FlowRunner account yet, [start free](https://app.flowrunner.ai). No credit card, and every integration is on every plan.

## Frequently asked questions

### What is Aster by AsterWise?

Aster is an LLM router with one OpenAI-compatible API. You choose one of two models, Aster Work for writing, research and everyday tasks or Aster Code for coding and repository work, and Aster picks the underlying model for each request.

### What can Aster do in a FlowRunner flow?

The Aster integration has two actions. Generate Text sends one prompt, with optional system instructions, and returns the answer as plain text with the finish reason and token usage. Create Chat Completion runs a full chat completion with conversation history, function tools, tool choice, reasoning effort, an output token limit and sampling controls.

### How do I connect Aster to FlowRunner?

Create an API key at app.asterwise.dev and add a balance there. Then paste the key into the Aster connection in FlowRunner. Aster usage is billed to your Aster balance.

### Does the Aster integration stream responses?

No. Each Aster step returns one response rather than streamed chunks. That does not guarantee a complete answer: the output token limit can cut it short, so check the finish reason. A finish reason of length means the answer was truncated.

By [Mark Piller](https://flowrunner.ai/about), Founder of FlowRunner [Editorial policy](https://flowrunner.ai/editorial-policy)

---
Markdown version of https://flowrunner.ai/blog/aster-llm-router-on-flowrunner. Site index: https://flowrunner.ai/llms.txt
