--- title: "Context API" id: "91598" type: "page" slug: "platform-enrichment-orchestration-context-api" published_at: "2026-06-08T17:40:36+00:00" modified_at: "2026-06-09T18:42:12+00:00" url: "https://tealium.com/platform-enrichment-orchestration-context-api/" markdown_url: "https://tealium.com/platform-enrichment-orchestration-context-api.md" --- Context API # Right Context for Every Agent, App, and Experience ## Tealium Context API delivers governed customer context from real-time profiles, warehouse data, and AI outputs to every agent, app, and experience that needs it. [Get a Demo](https://tealium.com/cdh-demo-2/) [Read the Developer Docs](https://docs.tealium.com/server-side/moments-api/) ##### The Shift ### AI Agents Need Context, Not Just Content. #### A model can't act on what it doesn't know. The agent does not see the last session. The recommender does not see the consent state. The service rep does not see the order history. ##### What Is It ### Context API Is Tealium's Governed Context Layer for the Agentic Era. #### Context API is a high-performance, customizable endpoint that returns only the customer context a specific agent, app, or workflow needs. It draws from real-time profiles in Tealium CDP, warehouse-modeled audiences from your data cloud, or AI-enriched outputs from any model. ## One API. Three Sources of Context. Context API serves what your business already knows from wherever it lives. ### Profiles, warehouses, and AI outputs through the same governed call. ### Real-Time Profile Context Read live customer state from Tealium CDP. Audiences, badges, attributes, and consent state, refreshed in the moment. ### Warehouse-Backed Context Expose modeled audiences and attributes from Snowflake, Databricks, BigQuery, or Redshift through the same low-latency API, with no second copy and no custom microservice. ### AI-Enriched Context Send live context to your models through Invoke Your Own Model, write predictions back to the profile, and serve the result through Context API for the next decision. ## How It Works Designed for Production AI ### 01 #### Build a Purpose-Built Endpoint for Each Use Case Define custom engines that expose only the audiences, badges, and attributes a use case needs, with up to 10 engines per profile by default. ### 02 #### Read Customer Context in Around 60ms Return small payloads in roughly 60ms so AI front doors, in-session experiences, and service workflows act before the moment passes. ### 03 #### Connect AI Agents Through Managed MCP Agent frameworks and LLMs query unified profiles, audience membership, and consent state through the Model Context Protocol without direct database access. ### 04 #### Serve Warehouse-Modeled Audiences as a Personalization API Pair Context API with Cloud Activation to deliver Snowflake, Databricks, BigQuery, or Redshift audiences through a real-time endpoint. ### 05 #### Turn AI Model Outputs Into Live Customer Context Score visitors with any model, write the result back to the profile, and surface it through Context API for the next agent, app, or workflow. ## Technical Detail Built for Production AI Enterprise-grade infrastructure designed for reliability, scale, and interoperability. | Latency | Around 60ms for small payloads. Vodafone benchmarks first-page reads at roughly 50ms in production. | | --- | --- | | Protocols | REST Managed MCP | | Engines per profile | Up to 10 custom endpoints by default. Each engine returns its own purpose-built payload. | | Session model | No active visitor session required. Useful for service workflows, agent calls, and offline contexts. | | Auth | API Key OAuth 2.0 | | Compliance | SOC 2 Type I & II · ISO 27001 & 27018 · HIPAA BAAs available · GDPR and CCPA support | ## Technical Detail What You Can Build With Context API ### Agentic AI Front Door Give every customer-facing agent governed access to live customer context through MCP, with consent and audience membership attached to every query. ### Service Assist Surface relevant order, audience, and intent context to contact center reps and chat agents the moment a customer connects. ### AI-Powered Product Discovery Feed real-time context to recommendation models and return personalized results inside the moment a shopper is browsing. ### First-Page Personalization Serve customer context in around 60ms for hero modules, offer logic, and onboarding flows. ### Warehouse-Native Personalization API Expose Snowflake, Databricks, or BigQuery audiences as a real-time API without standing up a microservice or copying data. ### Next Best Action and Decisioning Combine live profile state with model outputs to deliver next-best-action across web, app, call center, and agent surfaces. ## Built for Agentic AI ### AI agents need more than a prompt and an API key. They need customer context, consent, and an activation path. Context API gives them all three through one managed MCP server. Query Live Profiles Through MCP Agent frameworks pull profile attributes, audience membership, and consent state with a standard MCP call.Send Live Context to Models, Get Outputs Back Tealium routes the agent's context to your model, captures the response, writes it to the profile, and exposes the result through Context API for the next call.Respect Consent on Every Agent Action Every agent query carries the consent state of the user it affects. Agents can only act on data they are permitted to see.Stay Model Neutral Run agents on Bedrock, Vertex AI, OpenAI, SageMaker, or your own framework. ## Built for Enterprises. Proven by Results. ### [See more AI customer stories →](https://tealium.com/resources/?resource_type=case-study&sortdir=desc) ## 22% #### Real-time AI powered by Tealium ### Spark New Zealand Snowflake, and Spark's decisioning engine. Trigger to decision to activation in real-time. ##### [Read the case study →](https://tealium.com/resource/case-study/data-chaos-to-autonomous-marketing_spark-nz/) ## 54% #### More Call-to-Lead Conversion ### Legal & General Real-time identity resolution and activation drove measurable engagement lifts in a regulated financial-services environment. ##### [Read the case study →](https://tealium.com/resource/case-study/legal-general-transforms-customer-engagement-through-real-time-data-and-insight/) ## 50% #### Increase in Conversions ### With the right data signals, AI models, and orchestration, Vodafone turned predictive intelligence into measurable sales impact. ##### [Read the case study →](https://tealium.com/resource/case-study/vodafone-ai-powered-customer-revolution/) ## Why Context API Why Teams Pick Context API ### Profiles update after every event. Personalization APIs that depend on warehouse refresh cycles miss what just happened. ### One endpoint serves Tealium CDP profiles and Cloud Activation warehouse audiences. No two services to stitch together. ### Bedrock, Vertex AI, OpenAI, SageMaker, custom endpoints. Use what you have. Swap without rewriting. ### Up to 10 custom engines per profile so each agent, app, or workflow gets only the payload it needs. ### PII controls, policy-aware payloads, and audit trails are part of every Context API response, not bolted on. ## FAQ Common Questions ### How is Context API different from Moments API? Context API is the renamed Moments API with expanded scope. The product still serves real-time profile context. It now also serves warehouse-backed context through Cloud Activation and AI-enriched context through Invoke Your Own Model. The rename reflects the broader use cases. Existing Moments API integrations continue to work. ### Can we use Context API with our own AI model? Yes. Context API is model-neutral. Tealium feeds your models through Invoke Your Own Model, captures the output, and serves it through Context API. AWS Bedrock, Vertex AI, OpenAI, SageMaker, and custom endpoints are all supported. ### Do we need Tealium CDP to use Context API? Tealium CDP is the most common source of profile context. Context API also serves warehouse-backed context through Cloud Activation, which can run alongside or independent of Tealium CDP. ### What is the latency in production? Around 60ms for most existing payloads. Latency depends on payload size and engine configuration. Processing of newly collected data is dependent on the source and enrichment rules, and is separate from Context API delivery latency. ### Can our agent framework call Context API directly? Yes. Context API ships with a managed MCP server for agentic systems. Agent frameworks query live profile context, audience membership, and consent state through standard MCP calls. See AI + Your Data. ## See AI That Moves at Customer Speed ### Walk through a live customer-event-to-prediction-to-activation flow with a Tealium AI architect on your own stack. [Get a Demo](https://tealium.com/lp/demo-request/) [Lanch a Self-Guided Demo](https://tealium.com/tech-demos/)