The modern data stack lacks unstructured data processing capabilities, presenting significant challenges in connecting diverse data sources, developing and deploying deep learning models, and maintaining resilient, high-performing pipelines. 🔮 Instill Core addresses these issues by providing a scalable solution that integrates AI models with the modern data stack, offering end-to-end infrastructure tools that streamline the entire process. This makes AI accessible and effective for everyone, bridging the gap that previously required extensive in-house, non-scalable efforts.

🔮 Instill Core encompasses multiple subprojects that together form a full-stack AI solution, including

The main project is maintained in the instill-core repository.

The project also comes with a web-based UI application known as 📺 Instill Console, which offers a user-friendly, no-code and drag-and-drop pipeline builder for effortless pipeline creation and enhanced real-time and historical observability of pipeline runs. Users have the option to interact with 🔮 Instill Core through the ⌨️ Instill CLI and can integrate it using the 📦 Instill SDK.

#💧 Instill VDP

VDP stands for Versatile Data Pipeline. The foundation of 💧 Instill VDP lies in the concept of a highly flexible, extensible, and versatile pipeline that can handle a variety of data modalities by connecting and using various components (see ⚙️ Instill Component section).

The implementation is in the pipeline-backend repository.

#⚗️ Instill Model

Processing unstructured data is akin to a chicken-and-egg conundrum. AI models require data for training, fine-tuning, and evaluation, while the data processing itself necessitates effective AI models. These two elements are mutually dependent and often interwoven in real-world applications. ⚗️ Instill Model is responsible for managing MLOps/LLMOps services. Its role includes serving, fine-tuning, and monitoring models to ensure consistent performance in unstructured data ETL processes.

The implementation can be found in the model-backend repository.

#💾 Instill Artifact (Coming Soon)

💾 Instill Artifact orchestrates unstructured data to transform documents (e.g., HTML, PDF, CSV, PPTX, DOC), images (e.g., JPG, PNG, TIFF), audio (e.g., WAV, MP3 ) and video (e.g., MP4, MOV) into a unified AI-ready format. It ensures your data is clean, curated, and ready for extracting insights and building your Knowledge Base.

The implementation is in the artifact-backend repository.

#⚙️ Instill Component

In 💧 Instill VDP, a component serves as a basic unit for constructing a pipeline. The platform provides a diverse range of components, each designed for unique roles. These components enable pipelines to execute ETL (Extract, Transform, Load) operations on unstructured data. The available component types are:

The integration framework, written in Go, is housed in the component repository.

#📺 Instill Console

The console of 🔮 Instill Core is a web-based user interface designed to facilitate easy interaction with 🔮 Instill Core. It features a no-code, drag-and-drop pipeline builder, enabling even non-technical users to easily construct flexible VDP pipelines by merely dragging and dropping components. The emphasis is on simplicity and user-friendliness, aiming to offer a cohesive and intuitive user experience.

The console's maintenance is handled in the console repository.

#⌨️ Instill CLI (Coming Soon)

The command-line interface tool provides a direct and efficient way to interact with 🔮 Instill Core. It is designed for developers who prefer working in a terminal environment or need to automate tasks. The CLI allows users to manage and control VDP pipelines, components, and other resources without the need for a graphical interface. It is a powerful tool for scripting and automation, enabling developers to integrate 🔮 Instill Core into their existing development workflows.

The implementation is in the cli repository.

#📦 Instill SDK

The project provides a set of libraries designed to facilitate the integration of 🔮 Instill Core into other applications. It provides a programmatic interface to interact with the core services, allowing developers to build and manage VDP pipelines, components, and other resources directly from their code.

Currently, the SDK is available in two languages:

  • Python SDK: Ideal for data scientists and AI researchers, the Python SDK allows for seamless integration with popular data science tools and libraries. It is maintained in the python-sdk repository.

  • TypeScript SDK: Designed for web and Node.js developers, the TypeScript SDK provides type safety and autocompletion, enhancing developer productivity. It is maintained in the typescript-sdk repository.


🔮 Instill Core adopts an API-first design principle, enabling seamless integration with modern data stacks at any scale.

The API foundation relies on Protocol Buffers version 3 (proto3) as the Interface Definition Language (IDL) to define the API interface and the structure of the payload messages. The same interface definitions are used for both REST (via gRPC-Gateway) and RPC, allowing access to the API over different protocols:

  • JSON over HTTP
  • Protocol Buffers over gRPC

The interface definitions are maintained in protobufs with auto-generated Go codes in protogen-go and Python code in protogen-python. The official protobuf documentation can be found in our Buf Scheme Registry (BSR).

#Release stage

subprojects and components developed and maintained in 🔮 Instill Core use release stage defined as below to indicate their readiness:

AlphaAn alpha component indicates that it is under development, and Instill AI is actively collecting early feedback and issues reported by early adopters. Alpha components are not recommended for production use.
BetaA beta component is considered stable and reliable, with no further backwards incompatible changes expected. However, it may not have been tested by a large user base. Beta releases are intended to identify and fix any remaining issues before moving to the next stage.
Generally AvailableA generally available component has undergone thorough testing and is ready for use in production environments. Its documentation is considered sufficient to support widespread adoption.

Last updated: 6/10/2024, 1:16:31 PM