Sprint Demo · June 12–25, 2026

TechOps

Wanderu · TechOps Friday, June 26, 2026

Reliability · New service

Scheduled jobs that actually run — and tell us when they don't

  • Auto-discovers every repo's scheduled workflows and fires them on time
  • Surfaces drift — missed, late, or stuck runs at a glance, behind SSO
  • Self-heals stuck jobs and reconciles against GitHub every 15 minutes
  • Run any job on demand — from the dashboard or via MCP
Built for agents too LLM agents can review and monitor scheduled jobs as part of a regular workflow — not just humans clicking a dashboard.
GitHub Scheduler dashboard listing org-wide scheduled jobs with status and next-run times

Org-wide scheduled jobs — owner, status, drift, and one-click run.

Wanderu · TechOps

Data access · New

Query the warehouse from anywhere — even your phone

What changed

  • Before: a server running on your laptop, a per-laptop token, desktop only
  • Now: a Snowflake-managed server reachable from any claude.ai surface — web and mobile
  • Your identity, your access: each person signs in with their own Snowflake credentials
Want in? Request access to the Snowflake MCP in #ai — we set your default role and warehouse so it's read-only and cost-capped.
Claude mobile app answering a Snowflake query about today's purchases

Asking the warehouse from the Claude mobile app.

Wanderu · TechOps

Knowledge access · New

@Claude in Slack, grounded in our knowledge

What it does

  • Answers from Wayfinder — TechOps, hosting, and configuration knowledge, right in Slack
  • Picks up alerts in-thread: scopes blast radius, checks runbooks, isolates root cause
  • Acts, not just answers: queries Datadog, opens draft PRs, and flags #techops when a human is needed
  • No friction: no repo checkout, no per-user login, no leaving Slack
Claude triaging a Datadog WAPI 500 alert in a Slack thread

Triaging a Datadog WAPI alert in-thread — root cause, draft fix, and a heads-up to #techops.

Wanderu · TechOps

Infrastructure · Cost

Kafka, right-sized and cheaper

A sprint-sized investment upgrading our MSK clusters and cutting Kafka spend — capacity now matches real load instead of paying for headroom we never used.

What we did

  • Right-sized MSK clusters across dev, pilot, and prod
  • No downtime: MirrorMaker 2 moved topics across as services cut over
  • Removed unused clusters and cleaned up config drift along the way
  • Retired the oversized clusters once everything was cut over

Why it matters

  • Lower Kafka spend — the largest single cost lever this sprint
  • Cleaner, drift-free config across all three environments
Careful by design Caught a premature prod cutover, reverted it cleanly, and re-merged only behind an explicit approval gate.
Wanderu · TechOps