Getting Started

Before you start working with our APIs & SDKs, you must have the appropriate credentials as an approved 3rd-party API integrator in order to access the full set of functionality that Deep Labs' DeepDecision® platform offers. If you already have your credentials, just enter your client_id and client_secret to get started. If you don’t already have an account, contact us at [email protected].

These credentials provide different permissions, which determine what type of information can be shared with you.

Lifecycles

DeepDecision offers access to products and services via several different technical integration patterns ("lifecycles"). The following lifecycles are currently available:

Deep Topology Lifecycle

Deep Labs' Deep Topology Segmentation leverages a combination of customer and contextual data to create distinct and dynamic profiles of behavior. Millions of signals are leveraged to add additional meaning and improve the accuracy of predictions.

Basic Workflow

1. A set of needs are defined and training data is provided to DeepLabs
2. DeepLabs build a series of models based on the data provided.
3. The model is deployed and made available via our APIs 
4. A user sends a request for processing a profile via the API and receives a reference key.
5. A user retrieves all predictions and insights for a given set of transactions or profiles

Use Cases

DeepDecision leverages world state awareness and deep topology segmentation to address customer problems ranging from Account Takeover (ATO), Anti Money Laundering (AML), False Declines, Credit Applications, Account Onboarding, KYC, data enrichment, and Marketing Decisioning.

Example use cases include:

• Assessing accounts and identifying changing behavior that detects “drift” (from one Deep Topology shape to another), account takeover, or likelihood of higher spending (e.g., high customer lifetime value)
• Identifying potential sleeper cells or money mules
• Providing contextual insights for given geography at a specific time for data enrichment and alerts. 
• Proving deep customer analysis including psychographics and predicted responses to external stimuli.

Key Concepts

Profiles

A set of data about an individual user to generate insights and predictions.

Insights

A set of features generated from the inference dataset, primarily intended for visualization.

Predictions

A comprehensive set of Deep Topology signals generated from the inference dataset intended for direct use, or as input features for downstream models.

Reference

The reference is the unique identifier for the decision and is used to correlate all events for that decision. We use Universally Unique Identifiers (UUIDs) to ensure uniqueness across DeepDecision.

Decision Lifecycle

Leveraging DeepLab’s Deep Topology technology, the decision lifecycle provides a continuous score predicting risk, propensity consume, or similar problems. 

Basic Workflow

1. The problem is defined and labeled training data is provided to DeepLabs
2. DeepLabs build a Deep Toplogy-powered supervised model based on the data provided
3. The model is deployed and made available via our APIs
4. A series of predictions and insights are generated on the inference dataset using the previously fit model
5. A user sends event data via the API  and receives a reference key
6. A user retrieves all predictions and insights of interest using the provided reference key.

Use Cases

DeepDecision leverages Deep Topology dynamic risk & propensity profiles to address customer problems ranging from Account Takeover, AML, and False Declines to Credit Applications, Account Onboarding Risk, Marketing Decisioning, and Identity Verification for Two-Sided Marketplaces.

Example use cases for DeepDecision include:

• Identifying which accounts are most likely to file specious chargebacks (i.e., first-party fraud)
• Identifying which transactions should be approved or declined based on changing behavioral context
•  Verifying it's a real user trying to create an account or onboarding onto an application
• Verifying it's the correct user trying to login to an account
• Validating the correct user is using your platform/application
• Verifying that a user has a claimed device in their possession

Key Concepts

Events

An event is a collection of data related to a specific subject.

Examples of currently accepted events:

Decisions

For a given event Deep Decision will generate a continuous score based on the pre-built model for the client..

Reference

The reference is the unique identifier for the decision and is used to correlate all events for that decision. We use Universally Unique Identifiers (UUIDs) to ensure uniqueness across DeepDecision.

Contextual Lifecycle

 Leveraging millions of signals pulled in daily and covering the world,  the contextual lifecycle provide unique insight for a specific geography at a specific time to power AM/AI solution, power alerting services, and enrich data source to gain new insights. 

Basic Workflow 

1. User sends a request for a specific geographic and date and receives a reference key
2. The user pulls the resulting output using the reference key.

Use Cases

By overlaying DeepLabs contextual data clients’ data become more powerful, meaningful, and usable,  Use cases include forensic analysis, situational awareness, data augmentation for resale purposes, and insight discovery.   

Example use cases include:

• Providing contextual insights for given geography at a specific time for data enrichment for ML/AI solutions   
• Power alerts based on anomalies and severe events
• Enriching a dataset to provide more value when resold.
• Provide context for data solutions to enable better understanding and execution for users.  

Key Concepts

Contexts

Proprietary signals and signal families that Deep Labs uses to enrich the input data and extract additional insights (e.g., psychographics, geolocation, sentiment etc.)

Reference

The reference is the unique identifier for the decision and is used to correlate all events for that decision. We use Universally Unique Identifiers (UUIDs) to ensure uniqueness across DeepDecision.

Authentication

DeepDecision uses JSON Web Tokens (JWT) for authentication. When registering with Deep Labs for access to DeepDecision you will have been issued a client_id and client_secret.

If you have not been issued a client_id and client_secret please contact our support team at [email protected].

Terminology

Copyright © 2024 Deep Labs