This mode setting is a separate axis from Take Control — the stage-by-stage panel (Approve & Continue, Re-run Stage, Mark Complete, Rollback, Skip) that lets you drive an incident yourself. Take Control works identically in either mode; switching Supervised/Self-Driving does not turn that panel on or off. See Taking control mid-incident below.
The two modes at a glance
Self-Driving mode
In Self-Driving mode, the AI drives the incident through its full lifecycle without waiting for stage-by-stage approval:What you see in Self-Driving mode
- A live timeline as the AI moves through stages
- An email whenever an approval gate opens — sent to members with the
approvals:grantcapability who have opted in; this is opt-in and off by default per user, not a Slack DM or in-app notification (see Notifications & channels) - Full agent reasoning and confidence scores in the incident detail view
- A one-click “Take Control” button that’s always available
Auto-approve configuration
For teams that want fully autonomous operation on lower-severity incidents, Self-Driving mode supports optional auto-approve policies. These let specific action types execute without a human approval step. Two auto-approve knobs are available: Auto-execute low / medium-risk fixes (autonomous_execution, recommended) — Auto-executes recommendations when every proposed action is at or below a risk ceiling you set:
Because the ceiling caps at
medium, high-risk actions (rollbacks, config changes, scaling operations classified high-risk) are never auto-executed under this knob — they always fall through to a pause. That carve-out is specific to autonomous_execution; it does not apply model-wide. The separate auto_approve_hitl setting below bypasses risk-based gating entirely, including high-risk actions.
Auto-approve agent actions (auto_approve_hitl) — The most permissive option. Auto-approves every recommendation regardless of risk level, with no risk check applied at all — including actions the autonomous_execution ceiling above would never auto-execute. Two hard-coded exceptions remain regardless of this setting: P1/P2 severity incidents always require human approval (see the Warning above), and a recommendation that opens a fix PR against your GitOps repository always pauses for human review, even with this setting on. A fix PR mutates your source repo on merge, so the draft diff always gets a human look before it can be approved — neither carve-out can be bypassed. Outside of those two, auto_approve_hitl turns off risk-based gating entirely. Intended for fully autonomous low-stakes environments where speed matters most, short of P1/P2 incidents and code changes.
To configure, go to Settings → Organization Details. Auto-approve agent actions (auto_approve_hitl) appears under the Agent autonomy card once the org default mode is set to Self-Driving; Auto-execute low / medium-risk fixes and Maximum auto-executed risk (autonomous_execution / max_auto_risk) sit alongside it. Resolution follows an env-wins-over-org precedence, so a per-environment override does exist — it’s just not exposed in the UI today; it’s set via the environment’s settings JSONB / API only.
Supervised mode
In Supervised mode, sureops drives the incident pipeline the same way it does in Self-Driving mode — detection and triage run automatically with no gate — but every batch of diagnosis recommendations pauses for your explicit approval before resolution runs. Nothing executes until you approve, reject, or modify what the AI proposes.Supervised mode does not, by itself, pause the incident after every stage — that’s the separate Take Control axis described below. What Supervised mode gates automatically is the pre-resolution approval step above.
Taking direct control in Supervised mode
If you want to stop the incident after every stage and drive it yourself — with Approve & Continue, Re-run Stage, Stage Chat, Navigate, Mark Complete, and Release to AI — that’s Take Control, and it’s available identically in either mode; Supervised mode does not turn it on for you automatically. See Taking control mid-incident below.Approve & Continue
Run the next stage, then pause again when it completes.
Re-run Stage
Re-execute the current stage with additional context you provide.
Stage Chat
Ask the stage’s agent questions. It can query your observability tools in real time and reply with data.
Navigate
Jump forward to skip a stage, or roll back to revisit a previous one. Requires a reason.
Mark Complete
Mark the stage as done with notes — for when you’ve handled it yourself without agent involvement.
Release to AI
Hand control back to the AI. All context you’ve provided (notes, chat, re-run results) carries forward into the next agent run.
Auto-release timer
When you take control, a countdown timer starts (default: 30 minutes). This prevents an incident from being left indefinitely with a human holding control but no one actually driving it.- Extend using the icon-only extend button next to the timer (a Plus icon,
aria-label="Extend control by 15 minutes") — each click adds 15 minutes; there’s no UI control for a larger single extension - A warning appears 5 minutes before expiry
- On expiry, control reverts to the AI (“AI driving”) and a timeline entry records the auto-release — no email, Slack, or in-app notification is sent for this event
Taking control mid-incident
You can take control at any time — even while an agent is actively running. sureops uses cooperative cancellation: the agent finishes its current operation (an LLM call, a tool query) and then pauses. It never kills in-flight work. What to expect when you take control:1
Click 'Take Control'
The UI immediately shows a “handing over…” state. The agent flag is set.
2
Agent completes current operation
If an LLM call is in-flight, the UI shows “completing current operation.” This typically takes 5–30 seconds. The partial findings are preserved.
3
Control transfers
You see “You have control.” All agent findings to this point are visible and usable. The Stage Control panel appears.
Configuring the default mode
The default mode is resolved through a fixed precedence: stage → incident → environment → organization → system default. More specific settings override less specific ones; there is no per-severity rung in this resolution (see “P1/P2 protection” below).Org-level default
Settings → Organization Details, in the Agent autonomy card (“Org default · Agent autonomy”). Sets the baseline for all new incidents. All environments inherit this unless overridden.Per-environment override
Settings → Environment Details, in the This environment · Agent autonomy card. Useful when you want strict supervision in production but full autonomy in staging.P1/P2 protection is not a configurable override
There is no per-severity mode setting. Instead, P1 and P2 incidents get a hard-coded protection: auto-approve and autonomous-execution settings never apply to a P1 or P2 incident, regardless of mode or org configuration (see HITL Gates). This is a fixed safety gate, not a per-severity override table you configure.Per-incident override
The Agent autonomy control on the incident detail page overrides the environment default for that specific incident. The override is visible in the incident header and recorded in the audit trail.Comparison table
Trust progression
sureops is designed for deliberate trust calibration, not automatic escalation. Moving from Supervised to Self-Driving is always an explicit operator decision — the system never promotes itself to more autonomy. A common progression:- Week 1–2: All incidents run Supervised. Review agent reasoning, correct misclassifications.
- Week 3–4: Switch P4/P3 incidents to Self-Driving. Monitor audit logs.
- Month 2+: Enable Auto-execute low / medium-risk fixes (
autonomous_execution) for low-risk actions on P3/P4. Expand risk ceiling only after validating behavior. - Ongoing: P1/P2 incidents remain Supervised by default for most teams.