Uploading & Managing Your Data

Sequen develops models tailored to your proprietary catalog and user behavior. Uploading your data and establishing processes for real-time event ingestion and catalog updates are crucial for achieving optimal results when integrating with us. To power recommendations and other personalization capabilities, we ingest three core data entities: catalog, users, and user events.

Entities

Entity	Purpose	Typical Cadence
Catalog	Determines what can be shown or ranked	Batch on change (hourly–daily)
Users	Identifies who we are personalizing for	Upserts as profile changes
Events	Tracks how users interact	Streaming (near real-time)

Integration

Sequen will collaborate with you to prepare and ingest your data. We offer several methods for integration:

REST API
Data warehouse integration (S3, Snowflake, Databricks)
Custom connectors

Supported Field Types

Types are auto-detected on upload, and can be manually adjusted as needed. We support the following types:

Type	JSON shape	Example snippet	Notes / limits
id	”text”	`"id": "1349ADJASGRE"`	Primary key for entities.
string	”text”	`"sku_id": "SKU-123"`	UTF-8, ≤16 KB
number	123 or 45.6	`"price": 79.99`	64-bit float; integers are fine
boolean	true / false	`"available": true`	—
array<string>	[ “a”, “b”, … ]	`"category_ids": ["footwear", "sneakers"]`	Array of strings only (use an object for mixed types)
datetime	ISO-8601 string or Unix timestamp (number)	`"original_timestamp": "2025-07-10T14:09:12Z"` or `"original_timestamp": 1720618152`	ISO-8601 with millisecond precision or Unix timestamp in seconds or milliseconds
embedding	array<number>	`"embeddings": { "taste_v1": [-0.12, 0.09, …] }`	Vector of floats, ≤1,024 dims. Multiple named embeddings allowed inside the object.
relation	”text”	`"merchant_id": "merchant-1"`	Mapped to the Table that this field relates to.
object	{ … }	`"transaction": { "payment_method": "credit", "currency": "USD" }`	Nested JSON object, ≤4 KB

Data Selection

Sequen automatically explores your dataset and selects features most relevant to training your models. To achieve the highest performance, send us as many fields and datapoints as possible. Let’s get started by integrating your data catalog.

Introduction

Data Management v2

Search & Recommendations

Chat

Embeddings

Uploading & Managing Your Data

Entities

Integration

Supported Field Types

Data Selection