Scopix
Scopix · Visual intelligence infrastructure

Visual
Intelligence

Annotate, analyze, search, and reason over videos and images. Built for your professional standards.

[ Get started ]Read the docs

Most of the world's data isn't text. It's videos, frames, and images. Scopix makes sense of it all.

The platform

Pixels in. Data out.

End-to-end visual data infrastructure, from raw file ingestion to structured, queryable outputs at any scale.

01 · Ingest

Connect any source

Upload files directly, connect cloud storage, or submit via API. Images and video — any format, any resolution. No preprocessing, no conversion.

Incoming
PNGfloor-plan.png8.1 MB
MP4walkthrough.mp494 MB
JPGhero_banner.jpg4.2 MB
MOVdrone_survey.mov220 MB
02 · Process

AI-powered extraction

Vision model ensembles extract structured metadata, dense captions, OCR text, embeddings, and object-level features.

Extracted
description:"Two-story colonial with wrap-around porch, double-height foyer, …"
tags:["exterior", "residential", "colonial", "porch", "landscaped", …]
ocr_text:"123 Main St, Unit 4B — Listed 03/2026 — MLS #298174 …"
objects:[{"house", 0.97}, {"maple tree", 0.94}, {"sedan", 0.91}, …]
colors:[{"#8B7355", 0.34}, {"#4A7C59", 0.28}, {"#E8DCC8", 0.19}, …]
embedding:[0.0234, -0.1891, 0.4102, 0.0087, -0.3291, …]
03 · Deploy

SDK, API, and agent-ready

Structured outputs plug directly into your pipelines, agents, and internal tools. Python SDK, 70+ REST endpoints, typed execution contracts for complex multi-agent workflows.

scopix.py
from scopix import SyncScopix
client = SyncScopix(api_key=...)
# upload a video
video = client.files.upload(
file="site_walkthrough.mp4",
)
# find any moment by content
matches = client.agent_search.videos(
"forklift entering aisle 7"
)
Search

Search visual data the way humans do

Query by intent, not by tag. Find the right frame, region, or shot without hand-built taxonomies.

  • Pinpoints the exact frame, region, or moment — not just the file
  • Sort and reweight by what matters: meaning, content, layout, motion
See how search works
Color-based visual search results
Annotation

Every file becomes a queryable knowledge artifact

Vision model ensembles extracts structured fields, captions, regions, and embeddings on ingest. The output is typed, machine-readable, and ready for agent consumption. No glue code, no scraping, no fragile prompts.

  • 70+ field types out of the box, plus custom rules
  • Typed execution contracts compatible with agent frameworks
  • Provenance preserved: every field is traceable to its source region
Real-estate image with extracted compliance fields
Operations

Built for production volume

Scopix runs the same way for a thousand files as for a million. No manual sharding, no degraded latency under load.

  • Cloud-storage connectors and webhook-driven pipelines
  • Auditable usage and per-tenant access controls
  • Real-time sync across web app, SDK, and REST API
Production-scale file processing dashboard
Built for

Made for the way you work

Developers

A clean Python SDK and 70+ REST endpoints. Typed responses, idempotent writes, and a CLI you can pipe.

Data engineers

Pipelines, batches, and webhooks for keeping a queryable visual layer in sync with the rest of your warehouse.

Teams and professionals

A no-code web app for anyone who works with visual data. Upload, search, chat, and organize from your browser.

Three surfaces, one platform

Choose how you build

Same powerful primitives. Pick the surface that fits your team.

pip install scopix

Python SDK

Pip install and start extracting in minutes. Typed, async-first, and fits in a coffee break.

View SDK docs
70+ endpoints

REST API

Full programmatic control over ingest, annotation, and query. Every primitive surfaced.

API reference
No code required

Web App

Annotate, search, and chat over your visual data right from the browser.

Open dashboard

Start building today.

Free to try, transparent pricing, and a Python SDK that fits in a coffee break.

[ Get started ]Read the docs