The fix queue

A flag is useless without a fix. The fix queue is the system that turns "here's a problem" into "here's the work we did about it."

The lifecycle of a fix

Every flag moves through one of two lifecycles, depending on how you respond.

Lifecycle A — Quick fix

New flag → Triaged → In progress → Resolved

For most flags, this is the path. You see the flag, accept the suggested fix, click Mark in progress. You do the thing. You click Resolved. The agent re-verifies the underlying metric and closes the loop.

If the metric hasn't actually improved when you click Resolved, thola will say so:

"The flag was for runway < 3 months; current runway is 2.7 months. Closing anyway?"

You can override (sometimes the fix is in progress externally), but the gentle nudge prevents premature closure.

Lifecycle B — Playbook

New flag → Run Playbook → [steps...] → Resolved

For complex flags — the cost-reduction one, the conversion-fix one, the burnout one — the suggested fix is a Playbook: a multi-step guided process. See Playbook → Overview for how Playbooks work.

The Playbook lives inside the flag until it's done. Each step you complete updates the flag's status.

Owners

Every flag has a default owner. The defaults:

The default can be reassigned per flag. You can also set workspace-wide autopilot rules that route flags to specific people based on type — see Diagnostics overview.

When a flag is reassigned, the new owner gets a notification. The old owner does not get a "you've been removed" ping — too noisy.

The Risk Flags screen

Find the full queue at Risk Flags in the sidebar. The screen:

• Defaults to active flags assigned to you
• Has a tab for All active (every flag in the workspace, RBAC-scoped)
• Has a tab for Resolved (the audit trail)
• Has filters for severity, module, owner, age

You can multi-select flags and batch-assign them or run Quick fix on all of them.

Quick fix

For simple, deterministic flags, the Quick fix button does the whole thing in one click:

• An overdue invoice → drafts a reminder, asks you to confirm send
• A low-stock SKU → drafts a PO with the suggested vendor
• A misallocated lead → reassigns to a better-fit owner

Quick fix never auto-commits an external action. Even when it could; we deliberately leave the final tap with you.

Bulk fix

When you have many flags of the same type, the Bulk fix mode is useful:

1. Filter the queue to a single flag type (e.g. all overdue invoices).
2. Click Bulk fix.
3. thola shows you the action it would take for each flag side-by-side.
4. You select all / select some / deselect.
5. Confirm.

The bulk action is wrapped in a single task so you can audit (or undo) it as a unit.

The audit log of fixes

Every fix — quick, Playbook, or manual — is recorded. The Resolved tab shows:

• Which flag
• Who closed it
• How (quick fix, Playbook, manual)
• Which actions were taken (linked to chat / task)
• Whether the underlying metric actually improved

The "did the metric improve" column is the most valuable. If you see a fix-type that consistently resolves the flag but doesn't move the metric, the Playbook for that type needs work — and thola will surface that as a meta-recommendation.

Flag aging

A flag that sits unresolved for too long gets escalated:

• 🟢 Low → 🟡 Medium after 30 days unresolved
• 🟡 Medium → 🔴 High after 14 days unresolved
• 🔴 High → top of every dashboard after 7 days unresolved (it doesn't escalate further, it just gets harder to ignore)

You can dismiss a flag instead of escalating it — and dismissal with a reason ("this is intentional", "data is wrong") is preferred over silently letting things age.

Recurring fixes

Some fixes are not one-off. "Counter ops missed shift-end reconciliation" will happen again if you don't change the underlying habit. When you resolve a flag, thola asks:

"Want to turn this into a recurring check?"

If yes, you create a Playbook that runs on a schedule (daily, weekly, monthly) and produces a small task. Most of thola's most-loved features come from this pattern — solving a problem once and putting a watchdog in place.

→ Next: Playbook & workflows