01 · Stages 1–3
Build the evidence base
Pin an approved brand profile, confirm the campaign brief and review the entities and evidence the campaign may use.
The EntityStacks campaign workflow
From a proven brand entity to a published content stack.
Twelve controlled stages keep evidence, editorial decisions, generation and publication connected. Each approval advances one exact version; changing an earlier decision sends dependent work back for review.
02 · Stages 4–7
Design the entity stack
Shape keyword intent, topical coverage, planned assets and contextual links. Every link carries an editorial reason.
03 · Stages 8–9
Create and review
Authorize a specific BYOK model, generate one planned article at a time and edit the exact output before approval.
04 · Stages 10–12
Publish and observe
Approve an immutable S3 manifest, verify public content and optionally authorize one to thirty daily drift checks.
Twelve reviewable stages
Every stage has one job.
- 1
Evidence
Brief
Confirm the target, audience, topic and constraints.
- 2
Evidence
Research
Synthesize the approved brief and brand snapshot. This is not live web research.
- 3
Evidence
Entity review
Approve the brand facts and identity references for this campaign.
- 4
Entity stack
Keywords
Review clusters, intent and sourced metrics. Leave unknown metrics unknown.
- 5
Entity stack
Topical map
Define distinct topics and identify gaps in coverage.
- 6
Entity stack
Asset plan
Specify formats, page counts, destinations and organizational tier labels.
- 7
Entity stack
Link plan
Define source → destination links with an editorial reason for each.
- 8
Creation
Generation
Generate scoped drafts using approved inputs and an authorized provider allowance.
- 9
Creation
Content review
Review exact asset versions, evidence and quality checks.
- 10
Release
Publishing
Approve a publication manifest before any external changes.
- 11
Release
Verification
Verify reachability and links separately from index evidence.
- 12
Release
Monitoring
Run checks within the authorized cadence, URL allowance and expiry.
Approval mechanics
A revision changes the dependency chain.
Approvals are version-specific. Reopening research, entities or another upstream stage marks affected downstream work stale while retaining its history. Only one stage executes at a time for a campaign.
Approve exact output
The review records the stage, version, reviewer and approved output.
Invalidate dependencies
A reasoned revision reopens the stage and marks later decisions for review.
Authorize provider use
BYOK jobs pin the selected OpenRouter connection, model and allowance.
Keep tenant boundaries
Database functions and forced row-level security enforce workspace access.
AWS S3 publishing begins with a manifest.
An owner or admin chooses a saved AWS account, bucket and region after content review is approved. EntityStacks shows every final article URL and a sandboxed preview before publication can be authorized.
Reviewed article IDs must exactly match the approved asset plan.
Objects use a unique workspace, campaign and manifest path.
Conditional uploads prevent an existing object from being overwritten.
Successful uploads remain recorded when another article fails.
An uncertain upload is checked by content hash before a user-authorized resume.
EntityStacks never changes the bucket policy or applies a public ACL.
Verification
Fetch every public article and compare its complete content hash with the approved upload. A human approves the verification results.
Monitoring
After verification, an owner or admin can authorize daily checks for one to thirty days. The schedule has an explicit allowance and expiry.
Clear limits
Reachability and content matching do not prove indexing or rankings. Live web research, managed campaign checkout and automatic backlink creation remain unavailable.
Ready to map the first entity?
Create a brand profile, approve its source of truth, then open a campaign brief.
Create a workspace