Components

Build visually. Reuse Components. Extend with code.

Pipelogic Components are typed building blocks for models, streams, transformations, rules, sensors, APIs, and custom logic. Use them to assemble AI Backends visually, then extend the system with Python or C++ when your use case needs something custom.

Component Library

Start With Ready-to-Use Building Blocks.

Browse Components for inputs, models, transformations, integrations, and Applications. Combine them into AI Backends, inspect how they connect, and adapt them for your own systems.

Vision

RTSP Camera Input

Connect live camera streams to a Backend.

VisionCameraImageFrame

Vision

Object Detector

Detect objects in images or video frames.

VisionModelImageFrame→Detection[]

Rules

Zone Rules

Apply spatial rules to detections and trigger structured events.

RulesSafetyDetection[]→Event[]

Vision

OCR Extractor

Extract text from images, labels, documents, and screenshots.

VisionOCRImageFrame→TextBlock[]

Speech

Speech Transcriber

Convert audio input into structured transcript segments.

AudioSpeechAudioStream→TranscriptSegment[]

LLM

LLM Call

Send prompts, context, and structured inputs to an LLM.

LLMTextPrompt→ModelResponse

Inference

Triton Inference

Call NVIDIA Triton models from a Pipelogic Backend.

InferenceGPUModelInput→ModelOutput

APIs

Webhook Sink

Send Backend outputs to external systems via HTTP.

APIsIntegrationEvent→HTTP Request

Applications

Human Review Queue

Route uncertain or high-value events to human review.

ApplicationsWorkflowEvent[]→ReviewDecision

Soon: 300+ reusable Components for vision, audio, text, sensors, models, APIs, data processing, and business workflows.

Browse all Components

Visual Backend Builder

Drag Components into a Backend and connect the system visually.

Compose Components into AI Backends using the visual builder. Connect typed inputs and outputs, configure parameters, and inspect data flow before hitting production.

Compose without hiding the logic

The visual builder makes Backend logic easier to understand across engineering, product, and operations teams.

Configure without rewriting code

Tune parameters, connect streams, swap Components, and test changes without rebuilding the entire system.

Inspect before deployment

See how data moves from input to model, from model to rules, and from rules to Applications or integrations.

See the visual builder

Typed by design

Every Component has a clearly defined interface.

Components explicitly declare what they accept, emit, and require. This makes them incredibly easy to connect, inspect, and reuse.

Clear inputs and outputs
Each Component declares the stream types it consumes and produces.
Safer connections
Pipelogic can validate whether Components are compatible before deployment.
Configurable behavior
Expose parameters so teams can tune a Component without rewriting code.
Reusable across Backends
Use the same Component in many Applications, Backends, and deployment environments.

Object Detector

Pipelogic Component · v1.4.2

Inputs

ImageFrame

Code when you need it

Extend the platform by building your own custom Components.

Create custom Components in Python or C++ for proprietary models, internal services, specialized transformations, low-latency processing, or use cases that need deeper control.

detector.py

For fast iteration, AI libraries, and integrations. Python Components are ideal when you want to move quickly, connect AI libraries, call external services, or implement custom business logic.

Works with

PyTorchONNX RuntimeHugging FaceOpenCVUltralyticsLLMsAPIsData transformsBusiness rules

Create a Python Component

from pipelogic import Component, Input, Output
 
class ObjectDetector(Component):
    input  = Input(ImageFrame)
    output = Output([Detection])
 
    def process(self, frame: ImageFrame):
        results = self.model(frame.data)
        return [Detection(b, s) for b, s in results]

FAQ

Answers about Components, reuse, customization, and typed streams.

A typed building block that declares inputs, outputs, and runtime requirements so you can connect and reuse pieces safely.

No — assemble Components visually in the Backend builder. Code is only required when creating custom Components.

Inputs, models, rules, transforms, integrations, outputs, and more for vision, audio, text, sensors, APIs, and workflows.

Yes — use Python for fast iteration and AI libraries, or C++ for latency, throughput, and memory control.

Yes. Version, share, and reuse Components across Backends, Applications, and environments.

Types declare what a Component accepts and emits; Pipelogic validates connections before deployment to catch mismatches early.

Ready to Build?

Join technical teams using Pipelogic to build and deploy AI apps faster.