All features AI Actions

The assistant proposes. You approve. Then it runs.

The in-app assistant can now draft configuration changes and fixes as preview cards — each with a before/after diff. Rescan a domain, set a finding's status, accept a risk with an expiry, assign work, open a PSA ticket, stage a remediation. Nothing happens until a human clicks Apply.

Get started See a live demo
Previewbefore it runs
Everyaction audited
Youclick Apply
ai-actions Apply
RSCNRescan domainAPPLY
STATSet finding statusREVIEW
RISKAccept risk · expiryHOLD
DIFFBefore / after diffAPPLY
What it does

Preview, then apply

A card with a diff

Every proposed change renders as a preview card with a before/after diff — you see exactly what will change before it does.

A real action catalog

Rescan a domain, set a finding's status, accept a risk with an expiry, assign work to a tech, open a PSA ticket, or stage a remediation.

Per-tool guardrails

Off, Suggest only, Ask to confirm, or Automatic — set the autonomy level per tool in settings. Automatic is limited to low-risk actions.

Audited by default

Every action the assistant proposes or applies is written to the activity log — who, what, when, and the diff that was applied.

Per-client policy

Different clients, different autonomy

Let a trusted client auto-apply low-risk fixes while a sensitive client stays fully manual. Medium- and high-impact actions always require a human to confirm — there's no per-client override for that.

Trusted client

Opt a client in and low-risk fixes can apply automatically — the routine hygiene you'd have approved anyway, without the click.

Sensitive client

Keep everything manual. The assistant proposes, a human applies, every single time — no action slips through on its own.

Always-confirm floor

Medium- and high-impact actions require confirmation regardless of the policy. The safety floor can't be lifted per client.

Human in the loop

Nothing runs until you click Apply

The assistant is a drafting tool, not an autopilot. It can prepare the work — the decision to run it is always yours.

Propose, don't execute

The assistant assembles the change and shows the diff. It never reaches into a client's tenant on its own.

Guardrails you control

Per-tool autonomy — Off, Suggest only, Ask to confirm, or Automatic — lives in settings. Automatic is reserved for low-risk actions; raise or lower it any time.

Reversible and logged

Every applied action is recorded with who, what, when and the exact diff — so you can review, explain, or roll it back.

Part of one platform

Approve-first, end to end

Approve-First AI Actions extends SBCMSP’s remediation loop — every finding is ranked by real-world risk, paired with a proposed fix, previewed as a diff, applied on your click, and re-verified on the next scan.

  • Preview card with before/after diff
  • Per-tool guardrails in settings
  • Per-client autonomy policy
  • Every action audited & reversible
See the full platform
✓
Applied
on your approval
Explore more

Related capabilities

Active Defense

Turn a finding into a scoped, operator-approved action behind a guardrail spine that keeps every step reversible…

Learn more

Remediation Campaigns

Fix one recurring finding — or patch one vulnerable package — across every affected client as a single approve-first campaign…

Learn more

AI Remediation Guidance

Each finding pairs with AI-written, step-by-step remediation guidance — the fix an action can then stage for your approval…

Learn more

Run your first scan free

See a client’s real posture in minutes — then unlock all 1,692 checks.

Get started View pricing