Knowledge Automation¶

Kiket’s knowledge automation suite turns release telemetry, readiness signals, and project context into structured documents that are ready for review and publication.

Overview¶

Drafting service generates weekly status packets, release notes, and executive summaries aligned with the active project.
Meeting summariser ingests transcripts from stand-ups, planning reviews, or retrospectives and produces action-oriented summaries.
Review workflow queues every draft for human approval before it is published or delivered.
Delivery logs track where each document was sent (docs-site, email, Slack, or webhook) so you can audit communication history.

Feature availability: Knowledge Automation is gated behind the knowledge_automation feature flag and requires the Professional plan or higher.

Accessing Knowledge Automation¶

Open any project workspace and use the Knowledge button next to the dashboards link.
The Knowledge hub lists recent documents with their status, author, and project association.
Use the Draft Weekly Status form to generate a new document for a project, or jump into the command palette for other templates.
Paste or upload meeting transcripts using the Meeting & communication summaries panel to kick off automated summarisation.
Select a document to review its markdown, inspection metadata, and delivery history.

Drafting Workflow¶

Use the project command palette (Cmd/Ctrl + K) and run Draft weekly status packet, or trigger automation through custom recipes.
Drafts are created with a pending review status and retain the context snapshot that generated them.
Usage metrics are recorded for every draft; see the usage dashboard for consumption monitoring.

Review & Approval¶

Reviewers can approve or reject drafts directly from the document detail view.
Every review captures feedback, the reviewer, and timestamps while keeping an audit trail in knowledge_reviews.
Once approved, the drafting service can publish or deliver the document to the configured channels.

Delivery¶

Deliveries can be routed to docs-site pages, email digests, Slack channels, or webhooks.
Each delivery is logged in knowledge_delivery_logs with status, channel, and metadata for post-mortem or compliance.

Meeting & Communication Summaries¶

Upload a transcript (manual paste, Slack call export, or Zoom recording) from the Knowledge hub.
Per-organization redaction policies mask sensitive terms, domains, and emails before LLM prompts run.
Summaries include highlights, decisions, risks, and structured action items with owners and due dates.
Recipients that follow the project receive in-app, email, or Slack notifications once the summary is ready; throttling prevents duplicate pings.

Stakeholder Co-Pilot¶

Use the stakeholder co-pilot panel to generate executive briefs, customer updates, or audit packets using curated templates.
Select tone presets (concise, optimistic, formal, empathetic, candid) for persona-aware messaging and review past drafts in the approval backlog.
Confidence badges surface model certainty and feed directly into the approval queue alongside other knowledge automation outputs.
Professional plans receive three briefs per month; upgrading to Enterprise unlocks unlimited co-pilot drafts and removes quota prompts.
Access from project detail pages via the Knowledge tab for project-scoped drafts, transcript summaries, and quick co-pilot actions.

Evaluation Benchmarks¶

Use the managed knowledge benchmarking job (available to enterprise tenants) to generate accuracy/coverage/latency metrics using curated evaluation samples.
Target thresholds: coverage ≥ 85%, hallucination score ≥ 90%, action-item recall ≥ 80%.
Extend the sample file with scrubbed pilot data before uploading it through the benchmarking UI; archived datasets remain available in the knowledge evaluations workspace.

Performance & Load Testing¶

Trigger the transcript load-test harness from the admin automation console to simulate transcript throughput (stub summariser, no Vertex AI usage).
Results include total duration, transcripts per second, and average latency to confirm Sidekiq capacity.
For end-to-end checks, monitor Knowledge::TranscriptSummarizationJob queue depth and evaluation logs.

Runbooks & Compliance¶

Operational runbook: available in the internal enablement portal covering incidents, benchmarks, and training workflows.
DPIA annex: distributed through the compliance workspace so privacy reviewers can confirm controls and approvals.

Usage Instrumentation¶

Knowledge automation emits the following usage metrics:

Metric	Description
`knowledge.document_generated`	A draft was generated for a project/organization.
`knowledge.document_reviewed`	A reviewer approved or rejected a knowledge document.
`knowledge.document_delivery`	A knowledge document was delivered to a channel (docs-site, Slack, email, webhook).
`knowledge.transcript_ingested`	A meeting transcript was captured and queued for summarisation.
`knowledge.transcript_summarized`	A transcript produced a meeting summary knowledge document.
`knowledge.stakeholder_brief_generated`	A stakeholder co-pilot briefing (executive/customer/audit) was drafted.

Review these in the usage dashboard to monitor adoption and cost drivers.

Next Steps¶

Configure automation recipes to auto-draft release notes or executive briefings.
Extend delivery to your docs-site or knowledge base via the delivery service.
Combine knowledge automation with readiness signals to keep stakeholders informed automatically.