News
Kojable Improves Topic Discovery and Content Calendar Reliability
Kojable has improved reliability across the workflow connecting topic discovery, topic clusters, campaigns and content calendar generation.
The result is a safer and more dependable path from research to campaign planning, scheduling and content production. This is infrastructure supporting AI Answer Alignment, not a direct ranking or visibility change.
What changed
Completed discovery work is protected from later planning failures
Topic discovery and content planning are connected, but they are not the same operation.
- Discovery execution: identifying keywords, questions and related topic opportunities.
- Discovery output persistence: saving discovered keywords, clusters and supporting information.
- Post-discovery planning: turning approved topics into campaign sequences and calendar entries.
The updated behaviour recognises committed discovery output as complete. A subsequent planning issue remains visible, but it does not cause successful discovery work to repeat.
Answer: committed discovery remains completed when a later calendar-planning problem occurs. The planning issue retains its own status instead of turning durable research back into a failed discovery job.
Why this matters
Reliable AI Answer Alignment depends on a trustworthy path from research to action
Kojable helps companies understand how AI systems interpret their business, products, competitors and market.
Research becomes useful when teams can turn it into an organised programme of content, evidence and improvement.
That requires dependable links between:
- The workspace and original discovery run
- The keyword record and topic cluster
- The campaign and planned sequence
- The calendar entry and finished article
Reliable data lineage helps teams understand what was planned, why it was selected and which information gap the resulting content is intended to address.
More reliable topic discovery
Discovery success is no longer erased by a later failure
The reported incident distinguished completed work from a completed workflow. Discovery had identified keywords, created topic clusters, persisted its output and established the underlying research record before a later PostgreSQL verification problem affected planning.
With the released workflow:
- Discovery remains completed and its outputs authoritative.
- A later planning issue receives its own status.
- Completed research is not performed again unnecessarily.
- The interface distinguishes the part of the workflow that needs attention.
This reduces the risk of duplicate provider calls, repeated model work and conflicting discovery histories.
Safer PostgreSQL handling
Deterministic database errors are not treated as temporary failures
Advisory-lock encoding
A lock key could contain a null byte that PostgreSQL could not accept. The affected value is handled safely before the advisory-lock query.
UUID and text comparison
A post-discovery verification query could compare a UUID field with text using incompatible types. The comparison is corrected explicitly in the affected query without global casts, operators or broad type changes.
The deterministic PostgreSQL error codes involved are 22021 and 42883.
Within this affected workflow, those errors are classified as non-retryable because repeating an identical query cannot repair a structural encoding or type mismatch. This does not disable retries for PostgreSQL errors generally.
Clearer discovery-run authority
One completed run remains the source of truth
Retries can create several historical attempts for one discovery submission.
The workflow records:
- Which completed run is authoritative
- Which earlier attempts have been superseded
- Which durable job owns the final successful outcome
- Which historical runs remain available for auditing
Superseded runs are preserved rather than deleted, while one final completed run becomes authoritative for topic status, campaign planning and calendar generation.
Stronger keyword identity
Calendar planning does not depend on keyword text alone
Keyword text is not a safe primary identity. The same phrase can belong to different workspaces, discovery runs, clusters, campaigns or market contexts.
Every schedulable cluster member must preserve a stable, workspace-scoped keyword identity connecting the topic to the correct keyword record, topic cluster, discovery run, campaign and calendar entry.
When that identity cannot be confirmed, the sequence writer fails safely. It does not:
- Match by text alone or select an arbitrary keyword record
- Attach content to a different workspace
- Create a sequence from an ambiguous topic
- Create a calendar row whose lineage cannot be verified
This preserves the link between every planned article and the question, information gap, cluster and campaign that justified its selection.
Better Generate Calendar feedback
A successful request is not the same as a successful calendar operation
Reaching the backend does not prove that a sequence or calendar row was created. The response reports the real operational outcome, including:
- Sequence items planned and written
- Items skipped and their reasons
- Calendar rows created or reused
- The calendar-synchronisation outcome
Typed outcomes include:
KEYWORD_IDENTITY_MISSING- A planned cluster member lacks confirmed stable identity, so no ambiguous row is created.
NO_VALID_SEQUENCE- Planning produced no sequence item that was safe and valid to write.
CALENDAR_SYNC_FAILED- A sequence was created or resolved, but calendar synchronisation did not complete.
The frontend displays the actual planned, written, skipped and synchronised result. It does not automatically retry an action that requires a data or identity correction.
Safer remediation
Repair tools remain bounded and evidence-based
Identity issues can be repaired only when the correct record can be established unambiguously from existing, workspace-scoped evidence.
Remediation does not manually manufacture discovery jobs, topic clusters, campaign sequences, calendar rows or keyword relationships.
Benefits for Kojable users
A more dependable path from discovery to publication
Fewer duplicate runs
Committed discovery is not repeated because of a later planning issue.
More accurate statuses
Completed discovery, failed planning and calendar-generation problems remain distinct.
Dependable calendars
Calendar rows require valid, workspace-scoped topic identity.
Clearer feedback
Users can see why rows were written, reused, skipped or not synchronised.
Safer history
Completed runs and published-content records remain available for auditing.
Better lineage
Topics remain connected to the discovery, cluster and campaign that justified them.
What this means for AI Answer Alignment
Reliable content planning supports a clearer information environment
This release is not a direct ranking or visibility change.
It strengthens the infrastructure supporting AI Answer Alignment and answer-engine content planning by preserving relationships between research, approved topics, campaigns, scheduled content and finished articles.
A more coherent content programme can support clearer subject coverage, better-maintained company information and easier auditing of why content was created. It does not guarantee rankings, citations, traffic, recommendations or inclusion in generated answers.
What will not change
Safety and data-lineage requirements remain in place
- Workspace isolation and stable topic identity
- Valid cluster relationships and explicit campaign connections
- Safe sequence-writing behaviour and auditable discovery history
- Clear failure states and evidence before remediation
- Preservation of existing completed and published records
- Normal application workflows for jobs, sequences and calendar rows
Frequently asked questions
About the reliability update
Will the update delete previous discovery runs?
No. Historical attempts remain available for auditing, with one final completed run marked authoritative and earlier attempts explicitly superseded.
Will a calendar-planning failure repeat discovery?
No. Once discovery outputs are committed, a later planning failure remains a planning issue rather than rerunning completed research.
Can Kojable create a calendar entry from keyword text alone?
No. Schedulable topics require stable identity connected to the correct workspace and topic cluster.
What happens when Generate Calendar cannot create rows?
The response identifies missing keyword identity, no valid sequence or failed calendar synchronisation rather than reporting a misleading success.
Is the Generate Calendar workflow working in production now?
Yes. The approved production workflow preserves committed discovery, reports calendar outcomes and avoids automatic retries for deterministic data issues.
Will this release require a database migration?
No. The approved release required no new database migration.
Will this directly improve rankings or AI citations?
No. It supports reliable topic and content operations but does not guarantee rankings, citations, traffic or AI visibility.
How will this support AI Answer Alignment?
It preserves the relationship between identified information gaps, approved topics, campaigns and finished content, creating a more dependable planning process.