Arcus Documentation
Back to homepage

Key Concepts and Overview


Prompt enrichment provides external data and signals alongside your prompts for your Generative AI applications to boost the quality of their generated outputs. The external data used to enrich your prompts augments the Generative Model’s context with essential and up-to-date information that’s relevant to your task. Grounding your Generative AI application with external data using prompt enrichment prevents hallucinations, makes your generated outputs more precise and relevant and allows it to infer on live data. Arcus helps you discover and select which external data is most relevant to your application and helps you compose and build a custom blend of external data context that is specifically tailored to your application. This overall improves the performance of your generative outputs.

Data Augmented Generation

Arcus’ Prompt Enrichment offering leverages the Arcus Data Exchange to consume and inject data that transparently improves your generative models’ performance. The Arcus Data Exchange is a two-sided auction that connects your prompts to high-value external data.

You can get started by using the Arcus Prompt SDK, a Python library that integrates directly with common Large Language Model (LLM) providers and Diffusion Models and builds atop the exchange to provide prompt enrichment. Today, the Prompt SDK includes support for all OpenAI models, including GPT-4. It takes less than 10 lines of code to integrate your Generative AI workflow with Arcus.

Let’s establish the steps to integrate your Generative AI workflows with Arcus and consume high-value external data to improve their performance.

Prompt Enrichment Workflow

To use Arcus Prompt Enrichment, the first step is to create a prompt enrichment project on the Arcus platform.

When you create a new project, you’ll be asked to provide some basic information and requirements about your application. This information will be used to match your prompts to data candidates that are most likely to improve the Generative Model’s performance.

Once you’ve created a project, all it takes is to wrap your existing Generative AI workflows with the Arcus Prompt SDK and it will immediately start enriching your prompts with valuable context discovered by the exchange. Arcus supports a variety of Generative AI workflows, including:

  • Text Generation: Enriches your prompts for Large Language Models (LLMs) by integrating directly with your LLM provider to connect to the Arcus Data Exchange and inject valuable external data context directly into your prompts.
  • Image Generation: Enriches the prompts for Diffusion Models to make your generated images better grounded and more targeted.

How Prompt Enrichment Works

Arcus Prompt Enrichment uses Data Augmented Generation to provide Generative Models the context they need in order to give relevant and precise outputs. Arcus’ Data Augmented Generation works by taking your prompt to a Generative Model and augmenting it with additional context from relevant data sources, provided by the exchange. This consists of two main steps:

  1. Data Retrieval: The first step is to retrieve data from the exchange that is most relevant to your prompt and can provide important context to your generative model. This context can provide more targeted ground truth context to prevent hallucinations from the model, live or fresh data beyond what the underlying model has seen or more precise, targeted data for the generative model to infer on.
  2. Augmentation: Once the relevant data to your prompt has been identified and retrieved from the exchange, this data is composed as additional context to the generative model. For Large Language Models (LLMs), this is parsed and merged into the prompt and the LLM will leverage this context in addition to our original prompt to construct the generated output.

When provided with high value and relevant external data, Data Augmented Generation achieves state-of-the-art performance across both:

Arcus Prompt Enrichment supercharges Data Augmented Generation by connecting prompts to the Arcus Data Exchange. The exchange automatically matches your prompts to valuable external data to enrich the context provided to your generative models. This process leverages the Arcus Data Exchange’s powerful matching algorithms which rank external data candidates based on their inherent quality and relevance to your task to ensure that you’re always provided data that improves your generative model’s performance.


By using Arcus Prompt Enrichment, your Generative AI applications are no longer limited to your first-party data or the open domain data that they’ve been trained on. Instead, they leverage an intelligent ecosystem of external data to provide unique, essential and more targeted signals that are specifically relevant to your application.