[Analytics] Kinesis migration from Pinpoint: No event enrichment or session tracking

[sdorairaj-tab]

Is your feature request related to a problem? Please describe.

Summary

We are migrating from AWSPinpointAnalyticsPlugin to AmplifyKinesisClient (v2.54.0) as recommended by the Pinpoint end-of-support docs. The Kinesis client handles transport well (batching, caching, retries), but there is no replacement for the enrichment layer that Pinpoint provided automatically.

What's missing

When using AWSPinpointAnalyticsPlugin, every event was automatically enriched with:

• Device metadata (make, model, OS, locale, timezone)
• Application metadata (app ID, version, SDK version, Cognito pool ID)
• Client identity (persistent client ID, Cognito identity ID)
• Session lifecycle (start/stop/pause/resume tied to app foreground/background state)
• Structured event envelope (the JSON format delivered to S3 via Pinpoint event streaming)

AmplifyKinesisClient provides none of this. It accepts raw Data and sends it to a stream. Our downstream pipelines depend on the Pinpoint event envelope structure. Migrating currently requires manually reimplementing all of the above, including the session state machine (SessionClient.swift, PinpointSession.swift) from InternalAWSPinpoint.

Environment

Amplify Swift 2.54.1
Affects: iOS and Android

Describe the solution you'd like

Request

Provide a higher-level analytics plugin for Kinesis that replicates Pinpoint's event enrichment and session tracking. The logic already exists in InternalAWSPinpoint. Decoupling it from the Pinpoint API transport and making it usable with AmplifyKinesisClient would make the migration practical.

Describe alternatives you've considered

Manually reimplementing Pinpoint's device metadata collection, session state machine, endpoint tracking, and event envelope construction in our app's analytics provider. This works but is a significant effort for every Pinpoint customer migrating to Kinesis, and duplicates logic that already exists in InternalAWSPinpoint (SessionClient.swift, PinpointSession.swift). Ideally this logic would be decoupled from the Pinpoint API transport and made reusable with AmplifyKinesisClient.

Is the feature request related to any of the existing Amplify categories?

Analytics

Additional context

Related to Pinpoint - Kinesis migration

[ekjotmultani]

Hi @sdorairaj-tab, thank you for raising this issue, we are aware of these issues and have been actively discussing our way forward here, we understand kinesis really only solves one particular part of what pinpoint used to provide (sending events). When we have more details about our exact replacements or solutions we will return here with that information

[github-actions]

This has been identified as a feature request. If this feature is important to you, we strongly encourage you to give a 👍 reaction on the request. This helps us prioritize new features most important to you. Thank you!

[sdorairaj-tab]

Thanks for confirming this is being worked on. Two concerns:

Timeline - If a solution ships close to the October 30 deadline, we won't have enough runway. Any rough estimate on "weeks" or "months" would help us decide whether to wait, or pursue alternate solutions.

Extending Pinpoint Analytics support - Given there is currently no equivalent functionality in Kinesis for event details and session tracking, has there been any consideration to extending the Pinpoint Analytics API (putEvents, event streaming) beyond October 30, 2026 until a replacement is available?

We've been on Pinpoint since 2016 and our entire telemetry pipeline is built on it. This is a critical piece for our business. If it would be helpful for your product team to hear directly how this impacts our production app at scale, I'm happy to get on a call.

[mattcreaser]

Hi @sdorairaj-tab I've let our product team know about your request. We're working on some ideas about how to fill the missing gaps. I don't have any timeline information yet but it would be something we'd try to get out quickly to give you adequate time for migration.

Just to confirm: if you have something that enriches events in the ways highlighted above and then pipes those events through to Kinesis and/or Firehose, your existing telemetry pipeline could ingest the events from one of those sources?

[sdorairaj-tab]

Hi @mattcreaser. Thanks for following up. To answer your question, We'll still have to make some changes to our downstream pipeline - Pinpoint was adding an "arrival timestamp" field, which I guess isn't possible in a data stream service like Kinesis.

For the most part, having the other enriched fields would mean fewer changes to our pipelines.

[mattcreaser]

Thanks for the confirmation @sdorairaj-tab. We are looking into this and hope to have something to share soon.