Developer Hub

Real-Time User Context for
Agentic Apps & Analytics

Track behavioral events. Fetch live user context in <10ms. Build smarter AI agents and advanced analytics - all from one trusted data layer.

Try the Free Sandbox Explore the Docs

Spin up a testing environment for up to 7 days. Sign-up via GitHub - no credit card required.

Deliver Agentic Context to Your Favorite Frameworks

Snowplow combines live behavioral streams, delivering what users are doing right now in under 10ms, alongside historical context from your warehouse for a full picture of who your users are to serve them better experiences.

Get Started in a Few Easy Steps

Define the attributes

count_product_views = Attribute(
    name="count_product_views",
    type="int32",
    events=[Event(name="snowplow_ecommerce_action")],
    criteria=Criteria(
        all=[
            Criterion.eq(
                property=EventProperty(
                    vendor="com.snowplowanalytics.snowplow.ecommerce",
                    name="snowplow_ecommerce_action",
                    major_version=1,
                    path="type",
                ),
                value="product_view",
            )
        ]
    ),
    aggregation="counter",
)

Each attribute defines which event it will be calculated from, and what kind of aggregation will be performed.

Create an attribute group

attribute_group = StreamAttributeGroup(
    name="ecom_attributes",
    version=1,
    attribute_key=domain_userid,
    attributes=[
        count_product_views,
        count_add_to_cart,
        total_cart_value,
    ],
    owner="user@company.com",
)

Attribute groups organize related attributes together. They can be considered as "tables" of attributes.

Create a service

stream_service = Service(
    name="ecom_attributes",
    attribute_groups=[attribute_group],
    owner="user@company.com",
)

Services provide an interface for applications to retrieve attributes. Create a service that includes your attribute group.

Demo

4 mins

Build Real-Time Personalized Experiences with Context-Aware AI Agents

Watch Demo

Agentic Cookbooks

Experiment and learn with pre-built notebooks and demo applications across popular agent frameworks and technologies.

Python SDK Tracking for Ecommerce App

Enable agents to intervene before users have to explain themselves

~30 mins

Real-time Context for an OpenAI Travel Agent

Personalize content and chatbot responses based on user behavior

~45 mins

ML-based Prospect Scoring for a SaaS Website

Deliver real-time predictions to your browser for adaptive experiences

~45 mins

Own Your Analytics Stack, End to End

Snowplow delivers validated, event-level data straight to your warehouse. Model it with open-source dbt packages. Query it with standard SQL or a data agent - no black boxes.

Get Started in a Few Easy Steps

Track your events

// JavaScript tracker
snowplow('trackSelfDescribingEvent', {
  event: {
    schema: 'iglu:com.yourcompany/conversion/jsonschema/1-0-0',
    data: {
      conversion_type: 'signup',
      plan: 'pro'
    }
  }
});

Schemas enforce validation at ingestion. If an event doesn't match the schema, it fails — no silent data corruption.

Model your tables

# packages.yml
packages:
  - package: snowplow/snowplow_unified
    version: [">=0.5.0", "<2.0.0"]


dbt deps
dbt run --select snowplow_unified

Install Snowplow's Unified Digital dbt package to generate view, session, and user derived tables across web and mobile, all within your warehouse.

Query Behavioral Data

SELECT
    session_identifier,
    engaged_time_in_s,
    views_in_session,
    first_page_title
FROM <target_schema>_derived.snowplow_unified_sessions
ORDER BY start_tstamp DESC
LIMIT 10;