Introduction

What is Neuramancer?

We are a bavarian Start-up with an academical IT Security Background. Our multi-stage forensic analysis reveals image manipulation precisely and transparently, so that even partially manipulated images can be unmasked.

Our AI-powered image analysis API helps you detect deepfakes and manipulated images with high accuracy, providing detailed insights into potential alterations.

• Detect AI-generated or manipulated images
• Analyze compression artifacts
• Identify anomalies in digital images
• Generate heatmaps showing areas of interest
• Provide uncertainty metrics and confidence scores

API Overview

The Neuramancer API is a RESTful API. It accepts media payloads and returns detailed analysis results.

The API supports:

• Analyze Images: Our forensic Deepfake image analysis method doesn’t stop at REAL or FAKE decisions. We provide detailed prediction scores, uncertainties, multiple classes of influencing factors, such as compression artifacts and anomalies alongside visual 2D-heatmaps for every class. Currently JPEG and PNG only.
• Asynchronous processing: Create multiple analyses at once; poll for results or set a (presigned and frequently rotated) webhook for progress notifications.
• Custom Blob Storage integration: Seamlessly integrate with your existing blob storage solutions like S3 and Azure Blob Storage using our Blob Storage Management API.
• Webhook and API management: Use the API itself to rotate API keys and presigned webhooks based on your security policy requirements.

AI-Sovereignty and EU AI Act Compliance

Neuramancer is designed with data sovereignty and regulatory compliance in mind. We currently only host at selected ISO/IEC 27001:2022 certified, German datacenter locations. Should cloud hosting (from your perspective) not be an option, we also support:

• external media storage so you can host your media at the infrastructure you and we keep meta-data only or
• host our whole inference solution on-premise within your own infrastructure.

With this offering, we strive to meet the highest standards of regulatory requirements. On this mission, we are continuously working to ensure compliance with the EU AI Act and are currently in the process of obtaining an ISO/IEC 27001 certification.

Roadmap

We are actively working on new features and improvements for the Neuramancer API and SDKs. Some upcoming features include:

• Support for additional image formats
• Audio and Video analysis capabilities
• Multi-modal document analysis (e.g., PDFs, scanned documents)
• Real-time analysis via WebSocket API
• Enhanced reporting and visualization tools
• Additional SDKs (e.g., TypeScript/Node.js, Java, .NET)

Please contact our support team if you’re interested in specific features, would like to join our beta program, or have any suggestions.

Updates and Change Management

The Neuramancer API is under continuous development and currently operates in Beta mode. We regularly release new features, improvements, and bug fixes.

Release Notes Mailing

We will inform you about important updates via 📧 email if you have an active account with us.

API Update Methodology

Whenever we progress, our approach for all interfaces, including endpoints and data models is: IMMUTABLE-3-WAY-MERGE. We do NOT MUTATE endpoints, data models or behaviour.

There are three exceptions to this rule:

1. High severity security changes that require immediate action.
2. Legal or regulatory requirements that mandate specific changes.
3. Bugs that critically affect functionality.

In any usual case, we:

1. Add new endpoints, fields, data models, behaviours.
2. Deprecate old endpoints, fields, data models, behaviours (with advance notice).
3. Maintaining 100% compatibility with existing integrations until deprecated features are fully removed after the deprecation period ends.

Available SDKs

We provide official SDKs in two languages. Each SDK provides a complete client for the Neuramancer API with:

• Simple procedural API implementation of each individual endpoint
• Environment variable configuration
• Built-in validation and error handling
• Azure Blob Storage integration
• S3-compatible Blob Storage integration
• Command-line interface

Our SDKs focus on Simplicity and Security

Please note that we intentionally kept our SDK implementations extremely simple and straight-forward.

• There are no abstractions or complex design patterns involved.
• Each SDK ships a thin shared HTTP client (lib/http) and a config loader - every endpoint script builds on those two modules and nothing else.
• We don’t depend on any complex 3rd-party framework or library that might bloat the codebase or introduce security vulnerabilities (CVEs).

You may understand our SDKs as cheat-sheets or cookbooks. This deliberate design choice allows you to easily understand and adapt the code to your specific needs without unnecessary complexity.

PHP SDK

Best for: PHP web applications, WordPress plugins, Laravel projects, etc.

View PHP SDK Documentation =>

Python SDK

Best for: Data Science/AI/ML-teams and their Python applications, data science projects, ML pipelines, automation scripts.

View Python SDK Documentation =>