Faraday

Faraday AI agents for predicting customer behavior is a platform that equips data science and engineering teams to build proactive customer experiences using data, AI agents, and automation. It supports point-and-click workflows or API-based integration to deploy predictions across thousands of brands and platforms.

---

Overview

• Built-in consumer data with 1,500 attributes.
• Pre-packaged AI agents for key customer behaviors.
• Real-time and batch inference capabilities.
• Compliance and governance features including SOC-2 and CCPA considerations, along with bias management and explainability.
• Lifecycle support from data ingestion to deployment with integrated reporting.

How it works (high-level steps)

1. Connect data sources: Snowflake, BigQuery, Postgres, S3, or upload CSVs via API (POST /uploads).
2. Create Datasets: Map columns to identify people and extract the necessary events (POST /datasets).
3. Create Cohorts: Define key groups using event and trait artifacts (POST /cohorts).
4. Declare your prediction objective: Use built-in agents for core behaviors and set the outcome to predict (POST /outcomes).
5. Define a Scope: Choose population (Cohorts) and payload with the objective for deployment (POST /scopes).
6. Deploy with a Target: Add deployment destinations for your predictions (POST /targets).

AI agents and use cases

• Adaptive discounting: Estimate the value of promotional offers.
• Lead prioritization: Engage top leads first to maximize conversion.
• Next best offer: Predict what a lead or customer will buy next.
• Repeat purchase readiness: Identify customers most ready to buy again.
• Thematic personalization: Tailor messages and creatives to each target.

End-to-end quickstart (API-centric)

• Step 1: Connect your data sources (data warehouses, databases, cloud storage, or CSV uploads).
• Step 2: Create Datasets and map columns to identify people and events.
• Step 3: Create Cohorts to represent key groups.
• Step 4: Declare prediction objectives using built-in agents (outcomes).
• Step 5: Create a Scope for deployment readiness.
• Step 6: Create a Target to deploy predictions to your pipeline.

Ready-to-ship capabilities

• Built-in consumer data with extensive attributes.
• Pre-built AI agents for common customer behaviors.
• Bias management and responsible AI tooling.
• Real-time and batch inference.
• Easy deployment across multiple destinations and integrations.
• Privacy-conscious data handling and regulatory considerations.

Safety and governance

• Bias detection and mitigation tools.
• Explainability features to understand model predictions.
• Regulatory compliance support (SOC-2, CCPA, etc.).
• Clear data lineage from ingestion to deployment.

Core features

• Pre-built AI agents for key customer behaviors (e.g., lead scoring, purchase propensity, personalization cues)
• Seamless data source connectivity (Snowflake, BigQuery, Postgres, S3) and CSV uploads
• Dataset creation and column mapping
• Cohort creation and event/trait-based segmentation
• Objective declaration and outcome definitions
• Scope-based deployment planning
• Target deployment to multiple destinations and channels
• Real-time and batch inference
• Built-in bias management and explainability tools
• Compliance and governance capabilities (SOC-2, CCPA, etc.)