Skip to content
Product Module

Automated Drafts. Human Decisions. Nothing Publishes Without Approval.

Response Review is the reviewer dashboard for automated social response drafts: full source context, agent rationale, risk flags, and complete candidate account identity, with a signed, tamper-evident approval decision on every item.

Sample output
Decision Record
The exact approved text, the selected account, and the reviewer identity, recorded once and never overwritten.

Why It Matters

Results That Speak for Themselves

Human-in-the-loop
Nothing Publishes Without Explicit Review
Every drafted response lands in a reviewer queue with source context, agent rationale, risk flags, and the exact candidate account before a human decides. There is no autopilot path to publication.
Signed end to end
HMAC-Signed Contract in Both Directions
Staged drafts arrive over a signed request and approval decisions are returned over a signed callback. Replay and tampering are rejected at the signature layer before any business logic runs.
No overwrite path
Approved Text and Account Identity Cannot Change
Once a decision is recorded, there is no code path that can alter the approved text or substitute the publishing account. The identity a reviewer approves is the identity that publishes.

Before & After

Automated Drafting vs. Response Review

Automated response drafting is only safe to operate with a real human decision in the loop. Response Review makes that decision structural, not optional.

Without Aurora
  • An automated monitoring tool drafts a public response and you have no visibility into what it says before it goes out under your company's name.
  • You are not sure which account will actually post a given response, or whether the tool could quietly swap accounts after you approved something.
  • A reviewer edits a draft before approving it, and later nobody can prove what text was actually authorized.
With Aurora
  • Every draft lands in a review queue with the source post, the agent's rationale, and any risk flags. Nothing publishes until a person approves it.
  • The candidate account identity, including auth mode, connection details, and a verifiable identity hash, is shown next to the draft. That exact identity is what publishes; it cannot be substituted after approval.
  • The edited text becomes the approved text, hashed and stored at approval time. Any further edit requires a fresh review, not a silent overwrite.

Key capabilities

Everything a Reviewer Needs Before Approving Publication

Full identity context, risk flags, disclosure checks, and a hash-chained audit trail, all scoped to your organization.

Reviewer Queue With Expiry-Aware Sorting
Staged drafts are sorted by approval expiry, with a danger indicator under four hours remaining, so time-sensitive reviews are not missed.
Full Candidate Account Identity
Target account name, account type, user agent, connection source, egress type, auth mode and generation, every profile version, and a verifiable identity hash are shown, never collapsed to a bare account ID.
Edit-and-Approve With a Diff Indicator
Reviewers can edit the draft text before approving. A visible indicator shows when the text differs from what the agent originally wrote.
Risk Flag and Disclosure Checks
Risk flags render as a prominent banner, and the selected account's required disclosure language is checked against the draft text before approval.
HMAC-Signed Contract
Every staged draft and every approval decision is signed and timestamp-checked, with replay protection backed by a durable nonce store.
Hash-Chained Audit Trail
Every stage, view, and decision event is written to an append-only, hash-chained audit trail scoped to the review workflow.
What You Can Show Reviewers
Artifacts reviewers recognize, plus sample previews of structure.
Scroll for artifact previews

Connected to your stack

Connects to Your Stack

Pull evidence automatically from the systems your team already uses.

Common questions

Questions Teams Ask About Response Review

Security, integrity, and lifecycle answers for the Response Review module.

Can a response ever publish without a human approving it?
No. There is no code path in Response Review that publishes a draft without a recorded approval decision from an authenticated reviewer.
Can the account that publishes be different from the one I approved?
No. The identity hash submitted in the approval callback is looked up from the candidate the reviewer selected on the server; it is never accepted from client input. A forged or stale identity is rejected.
What happens if I edit the draft text before approving it?
The edited text becomes the approved text. It is hashed and stored at the moment of approval and there is no mutator that can change it afterward. Publishing a further edit requires a new review cycle.
What happens if the approval window expires before a decision is made?
The item is marked expired and a decision can no longer be submitted for it, regardless of what the interface displayed earlier. The source draft must be restaged.
How is the review queue protected from other organizations seeing our drafts?
Every read and write path is scoped to the reviewing organization at the database level, with dedicated automated tests that seed a cross-tenant record and assert it cannot be read.
Live walkthrough
Talk to Us About Response Review
Response Review is available as part of Aurora Command's automation surface. Talk to our team about connecting an automated drafting pipeline to a reviewer queue your team controls.
Share one request and we will show the path to decision record without losing approvals, ownership, or reviewer context.