Refactor Redis Agent Memory documentation and add SDK quickstarts

[raphaeldelio]

Summary

This PR restructures the Redis Agent Memory documentation around two reader goals:

1. Integrating applications with Redis Agent Memory under Develop with Redis.
2. Deploying and operating Redis Iris context engine services under Redis Products.

Documentation structure

The primary developer documentation lives under:

Develop with Redis → Redis for AI and search → Redis Iris context engine → Redis Agent Memory

Redis Iris is now treated as a separate product from Redis Cloud. Its operational documentation lives under:

Redis Products → Redis Iris context engine

This product section contains Redis Agent Memory, Redis Context Retriever, and LangCache. The pages explain how to create and manage these services on Redis Cloud. Self-managed Redis Agent Memory documentation remains under the Redis Agent Memory product section.

Previous Redis Cloud context-engine URLs redirect to the new Redis Iris locations.

Redis Agent Memory landing page

The landing page now provides:

1. A direct definition of Redis Agent Memory.
2. A capability overview covering session memory, automatic summarization, automatic long-term memory extraction, retrieval, multi-session recall, custom memory types, and sensitive-data exclusions.
3. A side-by-side explanation of session memory and long-term memory.
4. A travel planning example showing how session memory, summarization, long-term extraction, and custom memory types apply to one conversation.
5. A deployment selection between Redis Cloud and the Redis Software private preview.
6. Links to the Python SDK, TypeScript SDK, and REST API quickstarts.

The previous open-source Redis Agent Memory option was removed because there is no supported open-source Redis Agent Memory product.

Developer guide

The Developer guide provides:

1. A four-step integration workflow.
2. A comparison of the Python SDK, TypeScript SDK, and REST API.
3. Connection and authentication requirements.
4. Definitions for sessionId, actorId, ownerId, and memory IDs.
5. Session event and long-term memory field references.
6. Session retention and automatic summarization configuration, including a worked threshold example.
7. Automatic and direct long-term memory creation.
8. Semantic, keyword, and hybrid search guidance.
9. Custom memory types with a trip_preference example.
10. Sensitive-data exclusions, their scope, and their advisory nature.

The client-library guides for implementing memory directly with Redis were removed. The Developer guide now focuses on using Redis Agent Memory.

SDKs are listed before the REST API throughout the Redis Agent Memory landing page, Developer guide, references, and sidebar navigation.

Quickstarts

This PR adds three quickstarts that follow the same travel planning journey:

1. Python SDK quickstart using redis-agent-memory.
2. TypeScript SDK quickstart using @redis-iris/agent-memory.
3. REST API quickstart using curl.

Each quickstart guides the reader through:

1. Creating a custom Redis Agent Memory service with an extraction cadence of one minute, automatic summarization after six messages, and two recent messages retained in full.
2. Checking service health and connecting with the selected client.
3. Storing and retrieving a conversation with session memory.
4. Waiting for and searching automatically extracted long-term memories.
5. Adding enough conversation events to trigger automatic summarization, then retrieving the compacted session.
6. Extracting business-specific information with a trip_preference custom memory type.
7. Verifying that semantic exclusions guide automatic extraction away from specified sensitive information.

The steps explain what each feature does, what output to expect, and which background processes may require a short wait. Production notes distinguish the short extraction cadence and summarization thresholds used for the exercises from values suitable for an application workload.

Redis Iris service configuration

The create and manage service documentation now covers:

1. Configurable extraction cadence.
2. Automatic summarization settings and threshold behavior.
3. Custom memory types, extraction prompts, and structured fields.
4. Semantic exclusions for sensitive data, including their limitations and scope.

Shared setup instructions keep these values consistent across all three quickstarts.

Consolidation

The standalone API examples page was removed. Its useful material is now covered by the REST API quickstart and the Developer guide.

The older Redis Cloud REST quickstart was also removed in favor of the shared developer quickstart. Existing links now point to the new location, and an alias preserves the previous URL.

Redis Cloud prerequisites and service creation instructions were moved into shared embeds to reduce duplicated content.

Redis Iris product documentation

The Redis Iris operational hierarchy is now:

1. Redis Agent Memory
   1. Create service
   2. View service
   3. Self-managed deployment and operations
2. Redis Context Retriever
   1. Create service
   2. View service
   3. View admin keys
3. LangCache
   1. Create service
   2. Use LangCache
   3. View and edit cache
   4. Monitor cache

The self-managed Redis Agent Memory documentation moved from Develop with Redis to Redis Products → Redis Iris context engine → Redis Agent Memory → Self-managed.

Aliases preserve the previous Redis Cloud and self-managed URLs. The Redis Products page template was updated to support the additional navigation depth.

Naming and technical corrections

References were normalized to Redis Agent Memory across the AI, Redis Iris context engine, Google ADK, Redis Cloud, and self-managed documentation.

The description of automatic extraction now reflects the current product behavior. Redis Agent Memory processes session events asynchronously and stores relevant information as long-term memory without requiring applications to trigger the process.

This PR also corrects the Redis Context Retriever name in its limitations section and updates the AWS Marketplace listing to its current title.

---

Note

Low Risk
Documentation-only changes with redirects/aliases; no application code or runtime behavior is modified.

Overview
This PR restructures Redis Agent Memory documentation for integration vs. operations and standardizes naming to Redis Agent Memory across AI, Iris, ADK, and self-managed pages.

Develop path: The landing page is rewritten around capabilities, a two-tier memory model, and a travel-planning example; the open-source deployment option is removed. The developer guide is expanded into a full integration reference (workflow, clients, identifiers, session/long-term memory, summarization, custom types, sensitive-data exclusions). Three parallel quickstarts (Python, TypeScript, REST) replace the standalone API examples page and the old Redis Cloud REST quickstart; shared embeds dedupe prerequisites and service-creation steps.

Operate path: Redis Iris context engine operational docs move from /operate/rc/context-engine to /operate/iris, with aliases on moved pages. Self-managed Agent Memory docs relocate from develop to Redis Products → Iris → Agent Memory. Create/view service pages gain sensitive-data exclusions; links across LangCache and Context Retriever point at the new Iris URLs.

Other: OpenAPI description text is tweaked; Google ADK integration docs use the Redis Agent Memory Server naming consistently.

^{Reviewed by Cursor Bugbot for commit 1cbb67b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{Reviewed by Cursor Bugbot for commit 1cbb67b. Configure here.}

[cursor]

Orphaned agent memory get-started embed

Low Severity

rc-agent-memory-get-started.md was updated to point at the new Iris and REST quickstart URLs, but nothing in the repository references it after the Agent Memory landing page removed the multitabs embed that consumed it.

 

^{Reviewed by Cursor Bugbot for commit 1cbb67b. Configure here.}