Skip to content

Sprint - May 4 to May 17 #56537

Closed
Closed
Sprint - May 4 to May 17#56537
Labels
sprintSprint planning

Description

@pl
pl
opened on Apr 27, 2026

Sprint Planning

List of incidents + any action items that haven't been scheduled yet

Team sprint planning

For your team sprint planning copy this template into a comment below for each team.

# Team ___

**Support hero:** ___
**Off during the sprint:** ___

## Quarter goals

1. Quarter goals, status (red/yellow/green) and action points if yellow/red

## Retro

### Tasks

<!-- Grab the high and low priority items from last time and add whether that item was completed or not -->

## Plan

### High priority

-

### Low priority / side quests

-

### Are any other teams impacted by this plan? If so, tag them here

-

Activity

  1. pl
    added
    sprintSprint planning
    on Apr 27, 2026
  2. pl
    pinned this issue on Apr 27, 2026

  3. pl commented on Apr 27, 2026

    @pl
    pl
    on Apr 27, 2026
    ContributorAuthor

    Team Ingestion

    Retro

    Details

    High priority

    • Isolate ingestion kafka topics
      • 🟡 Break out the ingestion MSK cluster @pl
    • PersonHog
    • Expand first point of capture to support better processing tooling
    • Set ingestion SLAs and metrics and commit to them
    • Improve per-partition ingestion throughput
    • Deduplicator
    • Event Filtering
      • 🟡 Implement more efficient event filtering based on the RFC @pl
    • WarpStream
      • 🟢 Test capture-mirror in US without S3 Express One Zone @eli-r-ph
      • 🟡 Migrate ingestion outputs to WarpStream @jose-sequeira

    Incident follow-up

    • 🟡 Repartition Session Replay in prod-us, rename topics in all envs @z0br0wn
    • Error Tracking Ingestion
      • 🟡 Implement circuit breakers for error tracking @z0br0wn
      • 🟢 Tune the alerts to use ingestion lag @z0br0wn

    Support follow-up

    Side quests

    • 🟡 Move capture to the golden chart @eli-r-ph
    • 🟡 Move ingestion to the golden chart and rename the deployments @pl
    • 🟡 Move warpstream charts to the upgraded helm template @jose-sequeira
    • 🟢 Fix all health and cost warnings in warpstream @jose-sequeira
    • 🟡 Break out the heatmaps pipeline @pl

    OKRs

    • 🟢 Isolate ingestion kafka topics
    • 🟢 PersonHog
    • 🟢 Expand first point of capture to support better processing tooling
    • 🟢 Set ingestion SLAs and metrics and commit to them
    • 🟢 Improve per-partition ingestion throughput

    Projects

    High priority

    High priority

    • Isolate ingestion kafka topics
      • Break out the ingestion MSK cluster @pl
      • Decomission / downscale the old MSK cluster @pl
    • PersonHog
    • Expand first point of capture to support better processing tooling
    • Set ingestion SLAs and metrics and commit to them
      • Break out the heatmaps pipeline @pl
      • Break out the client warnings pipeline code @pl
    • Improve per-partition ingestion throughput
    • Deduplicator
    • Event Filtering
      • Implement more efficient event filtering based on the RFC @pl
    • WarpStream
    • Infrastructure cleanup
      • Move capture to the golden chart @eli-r-ph
      • Move ingestion to the golden chart and rename the deployments @pl
      • Move warpstream charts to the upgraded helm template @jose-sequeira

    Incident follow-up

    • Repartition Session Replay in prod-us, rename topics in all envs @z0br0wn
    • Error Tracking Ingestion
      • Implement circuit breakers for error tracking @z0br0wn

    Support follow-up

    • Investigate empty responses for persons queries @nickbest-ph

    Side quests

  4. lricoy commented on Apr 27, 2026

    @lricoy
    lricoy
    on Apr 27, 2026
    Member

    Team Web Analytics

    Legend
    ⚪ = Not Started 🟡 = In Progress 🟢 = Completed 🔴 = Won't complete 🔵 = Abandoned 🟣 = Ongoing

    Support hero:

    @jordanm-posthog

    Retro:

    Details

    🟢 @jordanm-posthog shipping the summary emails for bringing people back
    🟢 @jordanm-posthog roll out live view for everyone (shot out to the marketing/editorial team for pushing this)
    🟡 @jabahamondes drill down ad groups and ads and other utms that can be valuable
    🟡 @jabahamondes @lricoy try using the precomputation / reuse intermediate results
    🟢 @jabahamondes general performance improvements without precomputation
    🟢 @lricoy add bot analytics log integrations before release
    🟢 @lricoy figure out tracking for web analytics insights and related queries on mcp/posthog ai
    🟢 @arthurdedeus changed teams and is way happier now
    🟢 @arthurdedeus discovery calls on CSP pivot for customer analytics + RFC
    🟢 @arthurdedeus add usage metrics to the person's page

    Low priority

    🔵 @lricoy figure out if we can just ingest aggregated values instead of raw events/logs

    High priority

    @arthurdedeus user interviews with AE's and TAM's for the CSP pivot
    @lricoy write the docs for bot analytics current usage and ask people to try it out
    @lricoy add real time and bot analytics tooling to the MCP
    @lricoy figure out the increased CPU usage on sessions query and use Robbie's precomputation
    @lricoy move the traffic-sim to open source posthog repo
    @jabahamondes drill down ad groups and ads and other utms that can be valuable
    @jabahamondes @lricoy try using the precomputation / reuse intermediate results
    @jabahamondes add Marketing Analytics tooling to the MCP
    @jordanm-posthog auto resolution for health issues on the wizard
    @jordanm-posthog give some love to the heatmaps feedbacks

  5. marandaneto commented on Apr 28, 2026

    @marandaneto
    marandaneto
    on Apr 28, 2026
    Member

    Team Client Libraries

    Support hero: @ioannisj then @dustinbyrne
    OOO: @marandaneto 1 day

    Quarter goals

    • ⚪ Goal 1: What makes PostHog more tailored to how you actually use it
      • ⚪ Discover SDK + default gaps
      • ⚪ Proactive user research (interviews, surveys, mobile/backend focus)
    • 🟡 Goal 2: Agent-friendly SDKs
      • 🟡 RFC
      • TBD
    • 🟡 Goal 2: PostHog libraries vNext
      • ⚪ Offline-first / async-first posthog-js
      • 🟡 Distributed tracing headers + remote config support
        • 🟢 Client tracing headers sent on mobile
        • 🟢 Server-side SDKs supporting request contexts read tracing headers
        • 🟡 All server-side SDKs support request contexts where applicible
        • 🟡 All mobile and web clients support first class tracing headers
      • ⚪ More framework-specific wrappers
      • 🟡 Mobile session replay (triggers, min duration, RN/Flutter/macOS, perf)
      • 🟡 Mobile error tracking (iOS improvements, RN support & restructuring, breadcrumbs)
      • ⚪ Flutter desktop support (Windows/Linux)
    • 🟡 Goal 3: Maintaining the SDKs
      • ⚪ SDK test harness + CI integration
      • 🟢 SDK rotation ownership
      • 🟢 Config standardization (API key/token)
      • ⚪ Ingestion rate limiting support
      • 🟢 Unified remote config endpoint (/static/token/config)
      • 🟢 SDK design guidelines and best practices
    • 🟡 Goal 4: Log for Mobile
      • ⚪ Add support for collecting and sending app-side logs from Android
      • ⚪ Add support for collecting and sending app-side logs from iOS
      • ⚪ Add support for collecting and sending app-side logs from Flutter
      • 🟡 Add support for collecting and sending app-side logs from React Native

    Retro

    Details

    High priority

    Low priority / side quests

    Plan

    High priority

    • 🟡 Define tasks for Agent-friendly SDKs @marandaneto
      • ideally run agents to document public APIs and configurations that are not yet matching the implementation
    • 🟢 Migrate repos using labels to release SDKs off to changesets @marandaneto
    • 🟢 Migrate repos to release from the main branch only and protect releases from only the main branch @marandaneto
    • Add request lifecycle enrichment / middleware to Rails, PHP, .NET @dustinbyrne
      • Initial release includes first class support for tracing headers
    • RN error tracking support @ioannisj
    • Minimum recording duration for Flutter @ioannisj
    • Manual capture API for Logs - iOS @turnipdabeets

    Low priority / side quests

    🟢 =finished 🟡=in progress 🔴=won't finish ⚪=not started

  6. rossgray commented on Apr 29, 2026

    @rossgray
    rossgray
    on Apr 29, 2026
    Contributor

    Team Batch Exports

    Support hero: @tomasfarias / @rossgray

    Quarter goals

    • Objective 1: Unify batch exports and real-time destinations
    • Objective 2: Reduce pipeline architecture technical debt
    • Objective 3: Self-serve debugging

    Retro

    Details
    • CDP backfilling
      🟡 Review metrics and alarms
      ✅ new batch exports Grafana dashboard

    🟡 Split up S3 destination into supported destinations into:

    • S3-compatible (catch-all)
    • AWS S3
    • Cloudflare R2
    • GCS

    ⚪ Integration for AWS S3 with support for no-credentials authentication
    🟡 Integration for PostgreSQL

    ⚪ Resource based allocator + auto-scaler for batch export workers

    ✅ Add new snowflake inserted_at column

    ⚪ Explore retryable/non-retryable in response body of CDP API

    • So that batch export can retry the request

    🟡 File download batch export feature (new)

    ✅=finished 🟡=in progress 🔴=won't finish ⚪=not started

    High priority

    • CDP backfilling

      • Rollout as beta
      • Add usage tracking so we're aware when users start using it
      • Add docs to CDP pages

    • Split up S3 destination into supported destinations into:

      • S3-compatible (catch-all)
      • AWS S3
      • Cloudflare R2
      • GCS

    • Support new auth methods for S3 and GCS via Integration model

    • Integration for PostgreSQL

      • migration of existing config

    • File download batch export feature

    Low priority

    • Resource based allocator + auto-scaler for batch export workers

  7. MichaelKutsch-ph commented on Apr 29, 2026

    @MichaelKutsch-ph
    MichaelKutsch-ph
    on Apr 29, 2026
    Contributor

    Team Infra

    Support hero:

    Retro

    High Prio

    • ⚪ MSK with kafka 4 @MichaelKutsch-ph
      • dev first plus switch-over
      • identify things for better automation
      • prod-* too once everything is clear
    • 🟡 more network things (tailscale setup cleanup) @MichaelKutsch-ph
    • 🟡 smokescreen rollout @Daesgar + golden chart integration- almost complete, waiting on CDOP
    • 🟡 migrate more apps to The Application™
      • django stack
      • admin api
      • web and worker still outstanding, needs some work still
    • 🟡 Cilium: starting EU rollout @danielxnj - PGB, django apps missing, rest is in flight
    • 🟡 argo notifications: rollout to teams @Z3r0Sum - automation wrapper missing
    • 🟡 VictoriaMetrics:
      • 🟡 Upgrade VM to latest @k33lan
      • ⚪ Node optimization VM: @danielxnj
        • Move vm to own nodepools (noisy neighbour)
        • More beefy nodes?
        • Tackle noisy neighbour (guaranteed QoS)
      • ⚪ Analyze expensive queries @langesven

    side quest

    • ⚪ GitHub secret terraform provider, start using it again @-support-hero
    • ⚪ Alerting / Monitoring for our gh actions
    • ⚪ argoCD appset templating

    ✅=finished 🟡=in progress 🔴=won't finish ⚪=not started

    OKR Infra

    High Prio

    cloud platform

    • migrate more apps to The Application™
      • django stack
      • admin api
      • web and worker still outstanding, needs some work still
    • argo notifications: rollout to teams @Z3r0Sum - automation wrapper missing
    • dev-argoCD @Z3r0Sum
    • VictoriaMetrics:
      • Upgrade VM to latest @k33lan
      • Node optimization VM: @k33lan
        • Move vm to own nodepools (noisy neighbour)
        • More beefy nodes?
        • Tackle noisy neighbour (guaranteed QoS)

    cloud foundations

    side quest

    • agent-orchestrated infra creation @Z3r0Sum

  8. mariusandra commented on Apr 29, 2026

    @mariusandra
    mariusandra
    on Apr 29, 2026
    Collaborator

    Tools, Teams, Data

    Q2 2026 objectives

    Objective 0 – Make the MCP a first-class, competitive data interface (George, Marius)

    • 🟢 End-to-end querying workflows: Create, run, iterate on, and debug SQL/HogQL queries with strong error messages, metadata, and performance guidance.
    • 🟡 Data discovery & context: Explore connections, schemas, properties, and other metadata. Provide guidance on which connections/tables/etc to use.
      • todo: properties
    • 🟡 Cross-database querying: Seamless querying with HogQL across ClickHouse, Postgres, DuckDB, and other sources with consistent ergonomics
      • todo: test the dialects
    • 🟡 Visualization & reporting: Generate charts, choose appropriate visualizations, and create reusable outputs like dashboards and reports
      • todo: use the right viz
    • ⚪ Notebook & workflow integration: Full notebook support via MCP, including creation, editing, and async/background workflows
    • ⚪ Analysis & intelligence: Using notebooks, answer business questions, run statistical analyses, detect inconsistencies, and optimize queries beyond syntax

    Objective 1 - Build a BI experience that feels competitive and joyful (Marius)

    • 🟡 New chart types, prioritizing the most-requested visualizations
    • 🟡 SQL editor reliability and UX improvements
    • ⚪ Better variables, including multiselect, query-backed and time-based variables (e.g. for dashboards)
    • ⚪ Easier schema exploration and an alternate “no-sql” query building interface (manual controls + AI integration).
    • ⚪ Ability to click in and out of rows to explore deeper.

    Objective 2 - Make notebooks the default way to analyze data (Anna)

    • ⚪ Python notebooks for customers
    • ⚪ Core notebook UX and stability improvements
    • ⚪ Better integration between notebooks, HogQL, DuckDB, and external databases
    • 🟡 Initial collaboration improvements for notebooks (lock nodes if someone is editing, etc)
    • ⚪ A clearer default analysis flow centered on notebooks

    Objective 3 - Build a first-class querying and DuckDB experience (George)

    • 🟢 First-class DuckDB support (not treated as “just Postgres”)
    • 🟡 Better query errors so failures are understandable and actionable
    • ⚪ Improvements to the HogQL type system to support earlier errors
    • ⚪ Calculated properties and lambdas on top of HogQL tables
    • ⚪ Improved UX for connecting to databases

    Retro

    • 💚 MCP - much smarter as a result, but still work to do
    • 🟢 SQL editor to notebooks (Filters, AI chat?)
    • ⚪ Projections / schema editing
      • Got a direction with dw sources, will merge the two approaches this upcoming sprint.
    • ⚪ Notebook papercuts and vibeation @annaszell

    Plan

    Support rota: George then Marius

    High priority

    • The rest of the MCP stuff @georgemunyoro
    • Merge direct query and sync sources in the UI. Build versioned projections and schema editing on top. @mariusandra

    Low priority / side quests

    Are any other teams impacted by this plan? If so, tag them here

    • N/A

  9. frankh commented on Apr 29, 2026

    @frankh
    frankh
    on Apr 29, 2026
    Contributor

    Team Logs

    Write more bytes somehow

    Retro

    Details

    High priority

    • 🟢 PII scrubbing / server side settings/filters / ingestion pipeline @DanielVisca
    • 🟡 traces out for posthog internal use in prod @frankh
    • 🔴 finish up backups @frankh
      • rory moving everything to ec2, taking over ownership of cluster 👍
    • 🔴 column configurations in saved views @jonmcwest
      • switched to MCP stuff
    • 🟡 have real internal users for alerts @jonmcwest
    • 🟢 Add alerting to MCP, and get claude to create a bunch of alerts to stress test @DanielVisca

    Low priority / side quests

    • 🔴 Get metrics MVP running in dev

    High priority

    • traces out for posthog internal use in prod @frankh
    • 5000 alerts running with no performance issues @jonmcwest
    • service level rate limiting @DanielVisca
    • simple ingestion controls e.g. log level, or sampling, or both @DanielVisca
    • Alerts hooked up to signals @jonmcwest
    • improve traces MCP (feature flag it) @DanielVisca
    • cut over clickhouse clusters to ec2 ones @frankh
    • sort out ingestion autoscaling in prod-us @frankh

    Low priority / side quests

    • send billing metrics for retention

    🟢 =finished 🟡=in progress 🔴=won't finish ⚪=not started

  10. rorylshanks commented on Apr 29, 2026

    @rorylshanks
    rorylshanks
    on Apr 29, 2026
    Contributor

    Team ClickHouse

    AKA Team Read Fewer Bytes Please

    Retro

    High priority

    Low priority / side quests]

    🟢 =finished 🟡=in progress 🔴=won't finish ⚪=not started

  11. robbie-c commented on Apr 29, 2026

    @robbie-c
    robbie-c
    on Apr 29, 2026
    Member

    Team Query Performance

    Support hero: @robbie-c
    Off during the sprint: barbados offsite!

    Quarter goals

    1. Quarter goals, status (red/yellow/green) and action points if yellow/red

    Retro

    We had the Lisbon offsite, the hackathon was great, team got 🪓 d

    Tasks

    Plan

    High priority

    @andyzzhao

    • Add resource limits
    • Ingestion side for new properties table (roll out for team 2 and profile full text index)

    @mjwarren3

    • April query performance review
    • Approved plan on implementing our data retention policy

    @robbie-c

    • auto-autoresearch in local dev
    • ship some PRs from discovered optimisations
    • block untagged queries for team 2

    @aspicer

    • dmat PRs and deploy plan
    • monitor offset removal
    • misc investigating query performance low hanging fruit

    Low priority / side quests

  12. lucasheriques commented on Apr 29, 2026

    @lucasheriques
    lucasheriques
    on Apr 29, 2026
    Contributor

    Team Surveys

    Support hero: @lucasheriques

    Quarterly goals

    🟡 Goal 1: Surveys research & pricing exploration
    🟡 Goal 2: Surveys MCP & skills
    ⚪ Goal 3: MCP analytics

    Retro

    • Work on Surveys' skills to make it simpler for people to create them without leaving PostHog AI/Claude Code/Codex @lucasheriques 🟡
    • Have at least 5 calls during this sprint to discuss surveys' pricing Replaced by finishing up Survey Translations. Emails were sent, hoping for the calls next week. @lucasheriques ✅
    • Look into adoption by medium and large customers @fivestarspicy ✅

    High priority

    • Have at least 5 calls during this sprint to discuss surveys' pricing @lucasheriques
    • Add metrics for monitoring how people are creating surveys after adding MCP/Skills @lucasheriques
    • Finish survey translations for mobile (RN/iOS/Android) @lucasheriques

    Side quests

  13. meikelmosby commented on Apr 30, 2026

    @meikelmosby
    meikelmosby
    on Apr 30, 2026
    Collaborator

    Team Workflows

    Time Off

    Retro

    High priority

    • Handover Documentation / Session around everything Email @havenbarnes @meikelmosby @dmarchuk

    • Being able to create/debug Workflows through MCP @meikelmosby

    • Email delivery throttle @meikelmosby

    • "Wait until event" workflow action / conversion goal @dmarchuk

    • Push Notifications for android / ios @havenbarnes

      • Look into GDPR
      • Look into just storing device tokens in person props

    • fix rate limiter being way too tight @meikelmosby

    • alert customers when they get rate limited with in app notifs@meikelmosby

    Low priority

    • Reminder: cleanup for cyclotron migration @havenbarnes
      • spin down legacy workers once the final delayed workflows have executed
      • clean up legacy code as well once it's no longer in hot path

    This Sprint

    High priority

    • new consumer for CDP & stability work / scaling stuff @benjackwhite

    • wrap up incident @meikelmosby

      • ship replay capabilities for blocked runs @meikelmosby
      • publish public post mortem
      • follow up comms with impacted customers

    • alert customers when they get rate limited (PR) @meikelmosby

    • Move us over to only use job-queue v2 for workflows @meikelmosby

    • Ship Email delivery throttle (depending on job-queue-v2 work) @meikelmosby

      • dedicated email queue PR
      • ses rate limiter PR
      • email send retries PR

    • Being able to create/debug Workflows through MCP @meikelmosby

    • "Wait until event" workflow action / conversion goal for team2 @dmarchuk

    Low priority

    • emit email events and make them opt-in @meikelmosby
    • LLM workflow node @dmarchuk
    • Merge push only for team2

  14. vdekrijger commented on May 1, 2026

    @vdekrijger
    vdekrijger
    on May 1, 2026
    Contributor

    Team Analytics Platform

    Support hero: @vdekrijger
    Off during the sprint:

    Legend

    ⚪ = Not Started 🟡 = In Progress 🟢 = Completed 🔴 = Won't complete 🔵 = Abandoned 🟣 = Ongoing

    Quarter goals
    Dashboard stability and enhancements @MattPua
    • 🟡 New and expanded tests to prevent unexpected failures
    • 🟡 Collection of UI/UX changes that make working with dashboards a smoother experience
    • 🟡 Better controls and components beyond showing just graphs
    Improve platform stability and reliability @vdekrijger
    • 🟡 A clear definition of "success" for alert checks, subscriptions, exports, and query service
    • 🟡 Stability improvements to better deal with degraded infrastructure
    • 🟡 A 99.95% success rate for our background jobs over the last 28 days rolling
    • 🟡 Anomaly detection to proactively spot regressions
    AI-first support for our owned components @vdekrijger
    • 🟡 Signal integration for specific alerts
    • 🟡 AI suggestions during creation
    • 🟡 Improved AI onboarding
    • 🟡 Investigate ways to deal with adjusted usage patterns from AI

    Retro

    1. Loved:
      What went well during the project? These are the things that team members appreciated, enjoyed, or found to be effective.

    @vdekrijger - Happy that migrating Alerts from Celery to Temporal came with the expected stability increase, it's interesting to see how "noisy/vulnerable" Celery workers are to rollouts, vs Temporal which deals with this way better (partially due to our configuration I presume).

    @vdekrijger - Appreciate that @MattPua still jumped in on certain Slack topics even though he is on vacation 😄.

    1. Loathed:
      What made things worse? What was missing or needs improvement? This category identifies gaps, such as missing resources, skills, or support that would have made the sprint more successful.

    @vdekrijger - Nothing comes to mind apart from having lost a PostHog Code session due to a Cloud -> Local handover that failed, but notified the team of this and apart from that the app has been stable!

    1. Longed for:
      What do team members wish they’d had during the sprint? This could be anything that would have made the work more efficient or enjoyable, like better tools, clearer requirements, or more time.

    @vdekrijger - There are quite a few plates to spin due to the support work, but TBH it's a good signal for me to revamp and organise my todo list again 😄.

    1. Learned:
      What new knowledge or skills did the team gain? This could be technical knowledge, better communication methods, or insights into the project or process.

    @vdekrijger - Got to play with our local MCP setup, and I have to admit that it works great and it's trivial to test MCP changes locally now!

    Tasks

    @MattPua

    • 🟢 Finish work related to dashbboard templates for copying templates between projects
    • 🟢 Finish up new alerts chart from Lisbon Hackathon
    • 🟢 Finish setup for new machine
    • 🟡 New Dashboards for Alerts/Subscriptions Metrics

    @vdekrijger

    • 🟢 Migrate alerts from Celery to Temporal
    • 🟢 Tackle a few outstanding resiliency improvements for our owned components
    • 🟡 Investigate possibilities for adding SLOs for the query service
    • 🟢 Tidy up the hackathon into separate PR's
    • ⚪ Start on integrating AI as a first class citizen.
    • 🟢 Enhance our alert MCP tool to create slack / webhook destinations

    Plan

    High priority

    @MattPua

    • 🌴

    @vdekrijger

    • Add auto-disable mechanisms to the subscription flow so that we can better deal with misconfigured subscriptions (a la alerts).
    • Add additional tools to our MCP (e.g. one off variable overrides)
    • Catch up on outstanding issues / PR's in the TF repo
    • Continued support hero

    Low priority / side quests

    Are any other teams impacted by this plan? If so, tag them here

  15. gesh commented on May 2, 2026

    @gesh
    gesh
    on May 2, 2026
    Member

    Team Product Analytics

    Quarterly goals

    1. Make product analytics AI-ready

      • Clean charts in the component library
      • MCP works with insights
      • Investigate metric skill

    2. Long lasting quality improvements (mostly UI, UX, bugs)

    3. Keep building the things people need

      • Data warehouse in insights
      • Period over period comparison in funnels
      • Inline events in all charts

    Past sprint

    Details @thmsobrmlr
    • 🔴 Continue working on DWH insight papercuts, e.g.
      • Finish consistent implementation for coercing and validating data warehouse columns across insights
      • Move configuration to a second popover
    • 🟡 DWH support for retention insights
    • 🔴 Side quest: legacy removal
    • [unplanned] hotfix for issues with dashboard overrides and insight validation rules
    • [unplanned] looking into mcp gaps after company refocus

    @gesh

    • 🟢 Support two weeks
      • Open test accounts filter settings in the side panel
      • Fixed broken MCP UI apps after version version upgrade
      • Prevent series cohort filters leaking into breakdown cohorts for trends
      • Fixed insight flickering numbers on page refresh
      • Embedded insights not loading for some users
      • Do not crash when cohort breakdown limit is less than the cohorts length
      • Display cohort names in shared insights/dashboards
      • Align tooltip colors with the bars colors in trends
      • Count users correctly in retention with feature flag breakdown
    • 🟡 Continue with the MCP stuff
      - 🔴 Support more actors queries (Retention, Paths, Lifecycle)
      - 🟢 Better actors visualization in the UI apps (create separate viz, currently, we reuse the HogQL table view)
      - 🟢 Add recording links to the actors data
      - 🟢 Open actors as a posthog insight
      - 🟢 [Adhoc] Inline events support in trends tool

    @sampennington

    • 🟢 Iterate on the new editor sidepanel and remove old version
      - 🟢 Use space better on small screens
      - + when AI panel is open
    • 🟢 Complete new Line chart components
      • Testing
        • 🟢 Performance
        • 🟢 Snapshots
      • Switch existing LineGraphs over to use new one
        • 🟢 Release for team
        • 🟢 Release for team 2
    • 🟢 bonus task: investigate-metric skill

    Plan

    • @thmsobrmlr
      • Support (maybe look into papercuts and signals)
    • @sampennington
      • Finish BarChart Component and use it in Trends
      • PoC for Hog-Charts in MCP Apps, make a migration plan
    • @gesh
      • MCP: Support more actors queries
        • Lifecycle actors
        • Retention actors
        • Paths actors
        • Funnels actors
        • Stickiness actors
      • MCP: Inline events support for funnels queries
    • @mjwarren3
      • Growth review next week (adding MCP metrics to that)
      • Data retention policy / implementation plan PR
      • Wrap up activation work

  16. richardsolomou commented on May 4, 2026

    @richardsolomou
    richardsolomou
    on May 4, 2026
    Member

    Team LLM Analytics

    Time off

    • Company offsite (1 week)

    Quarter goals

    Goal 1: Agent-first LLM Analytics @richardsolomou @brandonleung @carlos-marchal-ph
    Goal 2: Proactive LLM Analytics @andrewm4894
    Goal 3: Full evals stack @andrewm4894
    Goal 4: Product glue @brandonleung @richardsolomou
    Goal 5: Reliability and performance @Radu-Raicea @carlos-marchal-ph
    Goal 6: Trace and session UI @richardsolomou
    Goal 7: Ingestion pipeline and migration @carlos-marchal-ph @Radu-Raicea

    Retro

    Details

    High priority

    Low priority / side quests

    Mid-sprint check-up

    Plan (W19, W20)

    High priority

    Low priority / side quests

  17. ablaszkiewicz commented on May 4, 2026

    @ablaszkiewicz
    ablaszkiewicz
    on May 4, 2026
    Contributor

    Team Error Tracking

    @hpouillot OOO 3 days

    Retro 🟢 🔴 🟡 🚫

    @hpouillot - code owners usage for assignment rules 🟡
    @hpouillot - breadcrumbs web SDK 🟢
    @hpouillot - architecture schema for cost analysis 🟡
    @hpouillot - stretch: react and node SDK for breadcrumbs 🔴

    @cat-ph - Remaining MCP Tools 🟡 (few ones left. Doing great)
    @cat-ph - Skills for the wizard inside context mill 🟡 (in progress)

    @fivestarspicy - send outreach to ET users 🟢
    @fivestarspicy - pitch with Christophe for issue rate limiting 🟢

    @ablaszkiewicz - issue migration sync, clickhouse union with dirty issues 🟢
    @ablaszkiewicz - add recommendations to center (alerts, weekly digest, ingestion issues, cross-sell) - decided we will not do weekly digest, cross-sell. I started with long running issues 🟡
    @ablaszkiewicz - issue rate limiting 🟡

    OKR

    Recommendations - Setup & Priorities
    Grouping - Grouping
    Agent experience - MCP & Wizard
    Pricing exploration - Value alignment
    Issue Report - Signal integration
    Search query - Migration & Performance

    High priority

    @cat-ph - finish everything related to MCP (remaining tools and skills).

    @fivestarspicy - mcp user interviews,
    @fivestarspicy - mcp metrics in dashboards.

    @ablaszkiewicz - project wide rate limiting,
    @ablaszkiewicz - per isse rate limiting,
    @ablaszkiewicz - figure out background refresh for recs.

  18. haacked commented on May 4, 2026

    @haacked
    haacked
    on May 4, 2026
    Contributor

    Team Feature Flags Platform

    Support hero:

    Off during the sprint: Nobody!

    Quarter goals

    Link to goals

    1. Isolated flags-specific infra 🟡
    2. Load testing framework ⚪
    3. Sub-100ms P99, consistent response 🟡
    4. Decouple flag evaluation from persons DB 🟡
    5. Delete all /decide code 🟢
    6. Split remote_config from /flags 🟢
    7. Get /local_evaluation in a good state 🟡
    Details ⚪ = Not Started 🟡 = In Progress 🟢 = Completed 🔴 = Won't complete 🔵 = Abandoned

    Retro

    Big sprint for HyperCache infrastructure and performance work. Patricio finished the /local_evaluation Rust migration to 100% rollout and removed the Django implementation, then propagated mixed targeting and the /flags/definitions endpoint switch across the server-side SDKs. Patricio discovered there's a different (legacy?) API endpoint for local evaluation that's still getting hit and is investigating. Matheus shipped a stack of perf wins: regex caching, etag on hypercache writes, eliminating the zrange hotspot, and richer timing instrumentation. Haack laid the groundwork for porting HyperCache building to Rust (warm-flags-cache CLI, flags-cache-jumphost deployment with IRSA, lifecycle crate migration), kicked off the work to take billing writes out of the request hot path, began cleaning up stale posthog_featureFlagsHashOverride rows, and absorbed a long tail of canary CI fixes and cargo upgrades on the side.

    Details

    @haacked

    Flags HyperCache port to Rust:

    Hash key override cleanup:

    Rate-limit test determinism:

    Canary CI (side quest):

    Cargo upgrades (side quest):

    Other small fixes:

    @matheus-vb

    Performance:

    Rate limiting:

    @patricio-posthog

    /local_evaluation Rust migration:

    Switch local evaluation endpoint to /flags/definitions (cross-SDK):

    Mixed targeting in local evaluation (cross-SDK):

    Other:

    Plan

    Project Board

    High priority

    @haacked

    @matheus-vb

    @patricio-posthog

    Side quests

  19. TueHaulund commented on May 4, 2026

    @TueHaulund
    TueHaulund
    on May 4, 2026
    Contributor

    Team Session Replay

    Support hero:
    @TueHaulund

    OKRs

    Details
    • 🧠 Goal 1: Nail single-session on-demand summaries
    • 🎬 Goal 2: Scale video export pipeline
    • 🏷️ Goal 3: Build session categorizer
    • 📡 Goal 4: Replay as a Signal source
    • 🖥️ Goal 5: Build new AI-native frontend for Replay
    • 🔌 Goal 6: Basic MCP for Session Replay
    • 💰 Goal 7: Business model for AI features
    • 🔒 Goal 8: Map out PII, data retention, and IP concerns
    • 🗂️ Goal 9: Set up data labelling system for replays

    Retro ✅ 🔴 🟡 🚫

    Details

    🟡 @fasyy612 Custom summarization preamble/prompt + custom tags

    ✅ @ksvat Become MCP master and get an overview of current MCP

    ✅ @nicowaltz Wrap up labelling suite initial version
    ✅ @nicowaltz Coordinate data training efforts

    🟡 @fivestarspicy FinOps - figure out cost per active minute
    🟡 @fivestarspicy Beta learnings doc (session summarization)

    ✅ @TueHaulund Implement continuously sweeping summarization job to even out rasterizer load
    ✅ @TueHaulund Clean up feature flags and discontinued summarization types

    High priority

    @TueHaulund Replay Vision PoC

    @ksvat MCP feature requests (list summaries, curating playlists, misc bugs)
    @ksvat Streamable MCP

    @fivestarspicy Growth Review
    @fivestarspicy User research + pricing
    @fivestarspicy Add MCP metrics to dashboard

    @nicowaltz Surfacing XGBoost model
    @nicowaltz AI data agreement amendments worked out

  20. estefaniarabadan commented on May 6, 2026

    @estefaniarabadan
    estefaniarabadan
    on May 6, 2026
    Contributor

    Team Warehouse Sources

    Support hero: @estefaniarabadan then @Gilbert09
    OOO:

    • Estefania [May 18th to May 20th]

    Quarter goals

    • 🟢 Goal 1: Improve pipeline stability and reliability (Estefania)
    • ⚪ Goal 2: Add warehouse data integrations with other PostHog products (Tom)
    • 🟢 Goal 3: Add more sources (Marcus)
    • 🟢 Goal 4: Improved source config (Daniel)
    • ⚪ Goal 5: Bonus objective
      • User error notifications (including disabled due to errors) (Anyone)
      • Activation, retention metrics (stretch goal) (Anna)

    Retro

    Details

    High priority

    • 🟢 [Marcus] mcp improvements
      • update backend to have correct openapi types
        • remove custom handlers in the mcp
      • allow fetching sync log jobs
      • added suggest source imports skill
      • allow using integrations via the mcp
    • :large_yellow_circle: [Marcus] release slack with webhooks
      • request full Slack scope set in all environments
      • align slack source channel listing with destination implementation
      • edit webhook inputs after source creation is now possible
      • pagination on the wizard and edit page
      • pretty much ready - main blocker is slack rate limiting when setting up a lot of channels, because get_schema is being called very often. will need to think about this, but probably caching is the right call
    • 🟢 [Marcus] show warning when selecting an incremental field without an index
      • merged in prod
    • 🟢 [Marcus] new sources
      • put up a pr for polar.sh
      • put up a pr for intercom
    • 🟢 [Marcus] fixed frontend bugs
      • correct oauth redirect
        • persist existing values
      • fix: show unreleased sources with notify me button
    • 🟢 [Tom] Get started on some Skills
      • Got these into prod!
    • 🔴 [Tom] Get back onto working on Warehouse => CDP integration
      • Did other stuff with sources instead - http lib, resumable sources, customer io webhooks etc
    • 🟢 [Estefania] Investigate alternatives to Kafka for PipelinesV3 and decide based on it
    • 🔴 [Estefania] Continue the work on writing to DuckDB/DuckLake
    • 🔴 [Daniel] Get Postgres CDC over the line: finalize prod tests and release to a few customers. If it goes well, release to general pop under a early Beta flag
      • Tests in prod ongoing, not ready for release yet
      • Docs are done, PR is open but waiting for release
    • :large_yellow_circle: [Daniel] Extend Postgres CDC to other compatible sources (Supabase mainly)
      • Tested with Supabase, it works. But did not extend Supabase yet
    • :large_green_circle: [Daniel][Stretch] Database sources: multi-database/schema support
      • Multi-schema is almost ready, should be done by eow. Multi-db is postponed

    Plan

    High priority

    • [Tom] Get back onto working on Warehouse => CDP integration
    • [Tom] Auto build a bunch of sources via PostHog Code
    • [Estefania] Support
    • [Estefania] Keep the rollout for pipelinesV3
    • [Daniel] Postgres CDC prod fixes + extend to Supabase
    • [Daniel] Extend multi-schema support for DB sources
    • [Marcus] new sources
    • [Marcus] release slack with webhooks

  21. jurajmajerik commented on May 7, 2026

    @jurajmajerik
    jurajmajerik
    on May 7, 2026
    Contributor

    Team Experiments 🧪

    Retro

    Anders

    • CUPED

      • Mean and funnel metrics supported. Will test on our account before rolling out more broadly
      • Not sure about ratio and retention metrics yet, more complex, will likely wait with those, much less frequently used

    • Feature flags

      • Added SDK harness test adapters for flags on python, node, browser, go and elixir
      • RFC in the making, but figured out it would be good to go alongside a concrete example, so have coded up a PR for how it will feel here feat(flags): v2 flags makeover test #56506

    • Performance and reliability

      • Focused on this this sprint
      • Improved reliability by reducing OOM + Julian’s fix on workers killed on deploy
      • Need a full week to see trend clearly, but so far it looks much better
      • We can also toggle on “sync” queries now via flag, but holding back to see the effects of the above fixes on async queries first
      • Primary metrics dashboard updated to be weekly, good trends on performance and reliability
      • Shipped a couple of PR’s to improve our error tracking, trend should be visible once we have a full week of data in

    • Learnings

      • Nothing new, but got some “wow” moments with the MCP and performance analysis. With good data quality, agents do this really well. We should strive even more to make sure we have a strong “UX” signal in our event data. It has improved, but we still have gaps there. That will help us to be more proactive.
      • Full list of PR’s here

    Marcel

    • Support

      • Created new issues on github for customer requests
      • Improving clarity on a bias-inducing experiment setup reported by a customer
        • 56659 - Show multiple variants row exposures table
        • 56697 - Warn user about bias-inducing setup
        • 56455 - Fix special cases of metric event validation

    • AI

      • 55277 - Harden metric validation especially for agent-driven requests
      • Switching to new approach:
        • 55910 - Implement new MCP and Skills foundation (focus lifecycle management and guidance)
      • Monitor our MCP/Skill rollout
        • 56214 - Improve MCP schemas for common pre-existing agent issues (results-get wrong params, experiment-timeseries-results, better guidance in 404)
        • 56829 - Improve agentic guidance for uneven split experiments
        • 56906 - Expose pause state to MCP (to be merged)
      • Some devex:
        • Grouping and show-all for phrocs 55058, 55174, 55203

    Rodrigo

    Juraj

    • Precomputation

      • Added the ability to test-run queries in both direct scan and precomputed modes -> verified the results match up
      • Monitored precomputation - results look really good: HeyGen/ElevenLabs/Gizmo have no more timeouts for over a week now. BUT: even precomputed queries sometimes take over 100s -> will be very important to tackle this on the UI side/manage expectations (Rodrigo’s goal)
      • Improved query debugging scene with organization MRR (to triage high prio customers), links to experiment, etc.
      • Tested MCP+Skills, full list of cases here: https://gist.github.com/jurajmajerik/a4470fbe279b5ffb24b2da91144dd265

    • AI summarization

      • Fixed some bugs AI summarization - on cache miss there were no results returned and saved metrics were not included at all.
      • Finished with evals for now

    • Papercuts/UX

      • Added team-level “only include completed conversion windows”
      • Added unarchive action for experiments
      • Consolidate actions for shared metrics + add delete option
      • …

    • Helping onboarding Ruby in London

    Sprint Goals

    Anders

    • Feature flags

      • Get feedback from feature flags team on RFC and make a decision on the path forward

    • CUPED rollout

      • Publish docs
      • Roll out to customers safely

    • Performance & reliability

      • Continue to monitor and be on top of things

    • Query builder refactor

      • Collaborate with Rodrigo to decide an “end state” and move forward with it

    • OFFSITE!!!

    Marcel

    • Since I had also support this week, want to have another week to actually focus on AI:
      • Debug & solidify skill selection
      • Harden lifecycle mgmt
      • Jurajs findings
      • Extend to further endpoints
      • Dedicated time watching session replays

    Rodrigo

    • DW

      • Final production testing and docs updates.
      • Demo experiment for local development

    • Product backend migration

      • Complete the migration of the remaining endpoints

    • Cleanup backlog

      • Exposure diagnostics tool
      • Metric reloading cleanup
      • Release Actors Query

    • 🏖️Offsite 😎

    • PTO on Labour Day ⚒️

    Juraj

    • Precomputation

      • Will pause working on it for now - ratio/retention metrics do not show up a lot in heaviest queries so far. Feels like premature to optimize them.

    • Signals

      • Create a signal when significance is reached

    • Papercuts/UX improvements

    • Offsite 🌴

  22. pawel-cebula commented on May 18, 2026

    @pawel-cebula
    pawel-cebula
    on May 18, 2026
    Contributor

    Team Billing

    Retro

    Retro items

    @minekansu

    • 🟡 MIS 2Q26 update - this is almost done, materialization is glitchy for adjusted revenue calculations so i want to double check everything to make sure numbers are accurate and up to date
    • ✅ techstars workshop
    • ✅ weekly revenue updates
    • 🟡 active user and org for product analytics, session replay, flags - started working on this but got distracted by other tasks
    • ✅ set up contract setup rotation
    • ✅ sales hiring model scenario for layered coverage

    @pawel-cebula

    • In-house credit ledger
      • 🟡 Core ledger implementation
        • New in the stack this sprint: #1881 (CreditLedgerService with FEFO consumption), #1882 (proportional consumption), #1885 (preserve rollover credit source ratios)
      • 🔴 Integrate into existing customer balance transactions to prepare for shadow ledger
        • Should get to it this week
      • 🔴 Start running shadow ledger
    • PostHog Code - remaining steps ahead of public beta launch
      • ✅ Add billing period and plan details to seats APIs - #1862
      • 🔴 Usage and spend dashboards support - blocked by adding PostHog Code to usage reports
      • ✅ Bulk-provision posthog_code alpha pro seats via mgmt command - #1871
    • 🔴 Logs pricing - 30 and 90 day retention
      • Chatted with Abe and we might pick up batch export fast runs first
    • 🟡 Investigate limits incorrectly toggling back after being removed
      • Tackled by @abhischekt - TBC if the fix worked but we do get 7x speed up in quota limiting runs as a bonus 🚀
    • Other (not planned initially)
      • Billing MCP tools / API exposure (so agents can answer billing usage and spend questions):
      • Revenue model / IWA fixes:
        • ✅ Forecast cross-Customer scoping for shared Stripe customer pairs - #1878
      • Stripe / tax:
        • ✅ Echo prorated tax_amounts on credit notes for Anrok-taxed invoices - #1869
      • Alerts / infra:
        • ✅ Enable graceful shutdown in prod-us (HPA flapping follow-up) - charts#10364 - workday comparison: HPA scale events 50/day → 14/day (-72%), connection cuts 53/day → 8/day (-85%)
        • ✅ Investigated dev billing BillingNoHealthyUpstreams / BillingDownstream5XX (Karpenter AWS AuthFailure) with cloud-foundations
        • 🟡 Spike alert email - per-product breakdown + deep link - #1874
      • Goodie bag campaign (Every.to partnership):
      • Misc:
      • Support & cross-team:
        • 🟡 Investigated and triaged duplicate customer_balance_transaction credit grants surfaced by Ele - generated cleanup list, planning to add idempotency_key for prepurchased credits

    @mcoll-posthog

    • 🟡 Revrec baseline and revenue discrepancies reduction
      • 🟡 Improved ledger explorer
    • 🟡 S3 parquet export
    • ✅ Admin email backfill

    @abhischekt

    Q2 2026 Objectives

    ✅=finished 🟡=in progress 🔴=won't finish ⚪=not started

    1. Complete the transition to the new revenue model (eng lead: @mcoll-posthog , revops lead: @minekansu)

      • 🟡 Initial implementation
      • Discrepancy hunting and fixing - all understood and fixed or explicitly accepted
      • All internal dashboards re-implemented on the new model
      • Investor reporting migrated
      • Model understood and trusted across the team
        • Finance signs off on model documentation
        • Zero revops escalations for routine ARR/MRR questions for 4 weeks after launch
        • Bonus: self-serve (finance can pull monthly data they need to close the books without billing/revops in the loop)
      • Stretch goal: revenue model data flowing into Campfire

    2. Migrate all credits to an in-house ledger (lead: @pawel-cebula)

      • 🟡 Implement a double-entry credit ledger
      • Deploy and run the shadow ledger
      • Fix unexpected discrepancies
      • Migrate from customer balance to the ledger

    3. Ship all expected pricing launches and changes (lead: @pawel-cebula)

      • PostHog Code
        • 🟡 Seat-based subscriptions (free and paid)
        • Seat-based subscriptions (free and multiple paid)
        • Overages and background agents
      • Storage and compute
        • Data warehouse
        • Endpoints
      • Workflows
        • Channels: Push, SMS
        • Per-workflow limit exclusion
      • Batch exports
        • Higher frequency exports
      • Logs
        • 30 and 90 day retention

    4. Reduce tech debt and improve developer experience (lead: @mcoll-posthog)

      • Improve async task observability and reliability
      • Upgrade to PostgreSQL 17/18
      • Switch from Nginx to Granian

    5. Complete the Salesforce org object rollout (lead: @abhischekt)

      • ✅ Deploy deduplication workflow and scripts to production
      • 🟡 Complete historical migration to custom org object
      • 🟡 Migrate all integrations and lead triggers to new object
      • 🟡 Confirm Vitally and Salesforce sync works under the new setup
      • ⚪ Document all scripts and workflows in PostHog-owned GitHub repo

    6. Build customer health score v1 in Vitally (lead: @minekansu)

      • Define "active user" per product
      • Define org-level "active customer" rollup logic
      • Validate definitions against known accounts (healthy, churned, at-risk)
      • Implement health scores in Vitally

    7. Implement fully automated outbound pipelines (lead: @abhischekt )

      • 🟡 Source and score additional leads (LinkedIn followers, Appfigures.com targets, Stripe Propensity Leads)
      • ✅ Generate outbound sequence content using LLM in Clay
      • 🟡 Sequence highest scoring targets automatically using Lemlist

    This sprint

    @minekansu

    • sales growth review
    • sales ops handoff planning
    • finish MIS and board meeting numbers
    • skills for automating weekly revenue updates and monthly growth reviews
    • review finance/compliance docs janani shared last week

    @pawel-cebula

    • In-house credit ledger
      • Core ledger implementation
      • Integrate into existing customer balance transactions to prepare for shadow ledger
      • Seed the ledger based on the new rev rec model
      • Start running shadow ledger
    • PostHog Code
      • Dashboard to monitor for correctness and progress
      • Improve admin UX around seats
      • Run migration for alpha customers on launch day (May 5th)
      • Implement support for multiple paid plans
    • Batch exports pricing changes - faster runs
    • Ship first billing MCP tools and skills

    @mcoll-posthog

    @abhischekt

    Time off:

  23. marandaneto
    unpinned this issue on May 26, 2026
  24. marandaneto
    closed this as completedon May 26, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    sprintSprint planning

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions