Question 1

Does Scout actually fix things?

Accepted Answer

No - Scout is a scout. It watches, diagnoses, and calls it in to the right human (in plain English, in Slack) with the fix steps. There are two paid tiers on top: Plus is the hosted observability layer (managed dashboard, Slack alerts, fleet view) and Autopilot adds the action arm that performs the safe plays plus Investigate, the interactive forensic REPL. That split is deliberate: it keeps Scout and Plus dead simple to security-review (read-only on Tableau, write-only to Slack) and lets you add execution when you trust it.

Question 2

Will Autopilot ever take an action I didn't approve?

Accepted Answer

Not unless you turned on Autopilot for that scope. Autopilot only fires on the actions and within the scopes you explicitly enabled - site, project, datasource tier, owner role. Anything outside, Scout just calls in to a human instead.

Question 3

How does TableauOps handle our Tableau credentials?

Accepted Answer

The hosted product (Plus and Autopilot) takes a Tableau Personal Access Token at signup, scoped to your site. The PAT is encrypted at rest with envelope encryption, scoped per-org, never logged, and rotateable from your dashboard. We never see the underlying database credentials behind your datasources, your warehouse creds, or any of your data - only the Tableau PAT, which you can revoke from Tableau itself at any time. If you'd rather not share a PAT with a hosted service at all, the free OSS Scout runs entirely in your own environment, holds the PAT locally, and never phones home - same triage agent, same diagnoses, just self-hosted.

Question 4

Does this work with Tableau Server, or just Cloud?

Accepted Answer

Both. Scout authenticates via Tableau Connected App (JWT Direct Trust), which works for Cloud and Server. Server installs need REST API enabled. Tableau Server customers behind a corporate firewall can run the free OSS Scout self-hosted; the hosted Plus and Autopilot tiers work for any Tableau site reachable from the public internet (which is the default for Tableau Cloud).

Question 5

What does HeatScore actually do that severity doesn't?

Accepted Answer

Severity tells you what broke. HeatScore tells you what it's about to cost you - combining time-to-view, datasource fan-out, user/department exposure, and trust history. A failure on a Tier-3 sandbox at 4pm scores 12. The same failure on the exec dashboard at 8:45am scores 95. HeatScore is included in Autopilot.

Question 6

Where does Scout live?

Accepted Answer

Two ways. The OSS core is self-hosted: it runs in your own environment on Python + SQLite, talking only to your Tableau site, Anthropic, and your Slack webhook - nothing phones home. Plus and Autopilot are hosted on tableauops.com - you connect your Tableau site at signup and we run the dashboard, the alerts, and (in Autopilot) the executor. Pick the model that matches your environment: hosted if you want it managed, OSS if you can't share a PAT with a hosted service.

Question 7

Can I use Scout without paying?

Accepted Answer

Yes. Scout is free and open source on GitHub under Apache-2.0. It includes RefreshWatchdog, triage playbooks, the incident FSM, plain-English diagnosis, the Slack call-in, and the append-only audit log. The two paid tiers add capability on top of that: Plus ($249/site/mo) is the hosted observability layer - dashboard, managed Slack/email alerts, cross-site fleet view, 180-day retention; Autopilot ($799/site/mo) is the full action arm with Investigate (interactive forensic REPL), the executor, HeatScore, and the rest of the integrations.

Question 8

What's the difference between Plus and Autopilot?

Accepted Answer

Plus is monitoring + alerting, hosted. We watch your Tableau site, render the dashboard, manage the Slack and email alerts, keep 180 days of history, and give you a fleet view across up to 5 sites. Read-only on Tableau - we never write back. Autopilot adds two paid capabilities on top: Investigate (interactive forensic REPL on a single failure - concurrent-failure correlation, owner blast radius, schedule drift, upstream/downstream analysis, with a Markdown case file per session) and the Autopilot executor that runs the safe plays at the autonomy level you set (re-embed creds, retry refreshes, fix schedules). Autopilot is the diagnostician with hands.

Question 9

Can I upgrade from Plus to Autopilot mid-pilot?

Accepted Answer

Yes. Email us and we prorate the difference for the rest of the billing period - no contract reset, no setup fee, your incident history and baselines carry over. Most teams that start on Plus move up within 60 days, usually after a recurring failure pattern proves Autopilot would have closed the loop overnight.

Question 10

How does Scout handle false positives?

Accepted Answer

Two ways. First, the Triage classifier filters out known-noisy patterns before opening an incident. Second, the trust-history input on HeatScore dampens the score for any asset where Scout previously cried wolf - so it earns its credibility back.

Start the day with answers, not alarms.

Scout Core is open source.

From silent to recon- and Autopilot when you're ready.

Severity is what broke. HeatScoreis what it's about to cost you.

The four-part loop: Baseline → Detect → Triage → Call it in.

Open source. Hosted. Or fully automated.

Things admins ask before turning Scout on.

Stop being the first responder. Be the second pair of eyes.