Turnstone · always-on watchdog

Three data platforms.
One watchdog. Always on.

Turnstone is a Slack-native AI agent that watches your Informatica ETL jobs, Databricks pipelines, and Qlik dashboards around the clock. It auto-restarts failed loads against policy, flags stale rows and freshness drift before downstream reports break, and posts a daily lineage-and-freshness digest to the team's channel.

Talk to the architect →See how it responds
Platforms watched
Informatica
Databricks
Qlik
# data-platforms · pinnedlive
  • TS
    Turnstone2026-08-14 · 08:00
    daily digest

    47 pipelines covered · 2 alerts overnight, both resolved · lineage digest posted below for # exec_dashboard → # uk_sales_history.

  • TS
    Turnstone2026-08-14 · 02:14
    informatica · wf_uk_sales_daily

    Workflow failed at 02:11 (transient DB lock, exit 14). Would restart with the same parameter bindings. On-call engineer approval required.

    Approve restartInvestigate
  • TS
    Turnstone2026-08-14 · 14:32
    qlik · freshness drift

    App exec_dashboard 4h stale — 2 dashboards downstream. Will hold next user-facing alert until digest window. Quiet hours: 22:00 – 07:00 (configurable).

    Open in Qlik
3 of 3 actions taken in the last 24h were approvedcadence: quiet hours respected · digest at 08:00

# data-platforms · live thread, not a mock once installed

Coverage

One agent. Three platforms. The ones behind your reports.

Turnstone speaks each platform's native API surface — PowerCenter workflow logs, Databricks jobs + table lineage, Qlik reload tasks — so it can act on what it sees, not just summarise it.

  • Informatica

    PowerCentre & IDMC workflows

    • Workflow + session run status (start, end, rows, exit code)
    • Parameterised restart with the same bindings the original run used
    • Daily lineage digest per workflow, not per row
  • Databricks

    Jobs, DLT pipelines, lakehouse tables

    • DLT expectations, streaming lag, and table freshness
    • Cluster restart + retry against the same job parameters
    • Stale-row detection before downstream notebooks pick it up
  • Qlik

    QlikView & QlikSense apps, reload tasks

    • Reload task exit + partial-reload flags
    • App distribution and bookmark health
    • Quiet-window alerting so the team wakes up to a digest, not a nudge

Response

How it responds — per platform, with the human still in the loop.

Every remediation action sits behind an explicit Slack approval. Auto-restart fails closed by default, surfaces the proposed diff, and hands the keyboard back to the on-call.

What Turnstone watches

  • Workflow + session run statusStart → end, rows in/out, exit code, parameter bindings
  • Mapping-level failuresSurfaced to Slack within 90 s, before the workflow itself emails
  • Parameter & integration-service healthDetects mis-bound parameter files and frozen connections

What Turnstone does

  • Parameterised restartSame bindings, identical integration service, scheduled at the next quiet window
  • Investigate-only modeMandatory for any workflow tagged high-risk (finance / regulatory)
  • Lineage digestOne message per workflow, not per row — readable in 20 s

Slack surface — Approval lands in-thread with the failed log attached as a snippet. On-call approves via Slack button or types /turnstone restart <id>.

J

James (technical architect)

Founder · Turnstone

Provenance

Built by someone who has operated these tools.

Turnstone comes from a technical architect whose career has spanned Informatica PowerCentre, QlikView and QlikSense, and now Databricks lakehouse engineering. The integrations are written by someone who has actually run these systems — pulled nightly ETL jobs into Qlik dashboards, debugged Databricks job failures at 2 a.m., and monkey-patched Informatica workflows under deadline — not just someone who has read about them.

That is also why Slack is the UX, not one of many destinations: features studied from a slide do not survive an on-call rotation.

Pricing

Per monitored pipeline — the unit that matters.

No per-row surcharge on top of growing Databricks workspaces. No per-seat fee that punishes your on-call rotation. Whatever you already spend on Informatica renewals and Qlik licences stays out of the formula.

Turnstone · pricing

One number. The number of pipelines you want covered.
  • Monitored pipeline countThe unit you already track in CMDB
  • Daily Slack digest (lineage + freshness)Included
  • Policy-driven auto-restart (human-approved)Included
  • /turnstone slash commands for ad-hoc statusIncluded
  • Per-row / per-seat / per-query surchargeNone — deliberate
Talk to the architect →Get a tailored quote

FAQ

Q&A, with the answer a CIO would actually want.

If your question isn't here, the architect answers mail at turnstone-4@polsia.app.

Get in touch

Bring your ugliest pipeline. We'll show you the digest.

A 30-minute walkthrough — usually with your production run history loaded. We walk you through what we'd watch, how the restart policy would read, and what the first morning-of digest would look like in your channel.

Talk to the architect →turnstone-4@polsia.app