Levin Riegner
L+R Newsletters
Review dashboard for newsletter construction, evidence, artifacts, and production runs.
Access is limited to approved Levin Riegner accounts.
Access Denied
This Google account is not on the allowlist.
Current Edition
The issue pack and preview currently exposed by the dashboard. Stories are global candidates; this tells you which generated edition the artifacts belong to.
Outputs
Latest generated artifacts for the selected newsletter. Use the preview to read the email-style HTML, and the issue pack to inspect the markdown source.
Needs Review
Stories flagged by the credibility pass. Open a story to inspect evidence, then record the editorial decision in the story drawer.
Newsletter Model
Configuration that defines the audience, cadence, approval gate, and publishing target for this newsletter.
Sections
Editorial buckets used to assemble the newsletter. The last column compares currently included story candidates to all known candidates for each section.
| Section | Description | Max Items | Included / Total |
|---|
Sources
Watched sources and source types. These are deterministic inputs checked by the research run before broader discovery is considered.
| Source | Type | Bucket | Health | Target |
|---|
Static Discovery Queries
Broad public-web searches configured in the package. These are static queries, not derived from emergent stories.
| Query | Status | Priority | From |
|---|
Google Alerts
Gmail alert subjects configured for this newsletter package, using the account and sender query from the runtime environment.
| Subject | Status | Provider | Window |
|---|
Design Resources
Evergreen UX references that are rendered deterministically instead of competing with news stories.
| Resource | Source | Description |
|---|
Cadence Jobs
Configured pipeline jobs, their schedules, and the steps they execute.
| Job | Stage | Status | Cron | Steps |
|---|
Issue History
Published or generated newsletter editions over time. Repeated generations of the same edition are grouped as builds under one issue date.
| Issue | Latest Build | Status | Stories | Artifacts |
|---|
Stories
All generated story candidates for this newsletter. Status comes from the pipeline; Decision is the human editorial layer.
| Story | Section | Status | Decision | Actions | Priority | Updated |
|---|
Query Performance
How much each research query contributed to discovered and included story clusters.
| Query | Discoveries | Clusters | Included | Yield |
|---|
Source Health
Recent status for configured sources. Failures here usually explain thin or stale research output.
| Source | Status | Changed | Failures | Checked |
|---|
Run History
Recent automated jobs for this newsletter, shown with human labels instead of internal job keys.
| Started | Job | Stage | Status | Metrics |
|---|
Audit Log
Recent authenticated dashboard write actions for this newsletter.
| Time | Actor | Action | Story | Change |
|---|
Internal Workflow
How the system turns configured research inputs into a draft issue preview.
Collect and Discover
Configured sources, static discovery queries, and Gmail Alerts produce raw discoveries with source URLs, titles, snippets, and timestamps.
Cluster Signals
Related discoveries are deduplicated into story clusters so repeated coverage strengthens one candidate instead of creating duplicate newsletter items.
Research Stories
Clusters become story rows with section, summary, why-it-matters text, priority, evidence links, and inclusion state.
Check Credibility
The credibility pass flags weak sourcing, rumor-heavy items, noisy pages, or stories that need human review before they shape an issue.
Build the Issue
The issue pack selects included stories, groups them into configured sections, and records the source material behind the generated edition.
Render and Edit
The renderer applies section-specific visual rules, then the optional LLM editor improves L+R voice while preserving existing links and section structure.
Validate and Review
Link validation runs after rendering. The dashboard exposes the preview, issue pack, story decisions, run history, and audit log for human review.
Human Editor Workflow
How an editor should use the dashboard to shape a strong issue.
Start with open questions
Use Overview → Needs Review for credibility flags, then open each story to inspect evidence, source quality, and query provenance.
Shape the candidate set
In Stories, mark each item as Accepted, Needs Rewrite, Parked, Rejected, or Pinned. Use notes to explain the editorial reason, missing angle, or rewrite request.
Balance the issue
Use Construction to understand section limits and source coverage. Use Diagnostics when a section feels thin, stale, repetitive, or too dependent on one source type.
Generate a fresh build
Click Generate new edition after decisions are saved. This builds a new issue pack, renders the Beehiiv preview, applies the LLM editorial pass, and validates links.
Read like the recipient
Open Latest preview and compare the preview diff. Check section rhythm, duplicated claims, weak implications, broken style, and whether links still support the story.
Iterate deliberately
If the draft is not ready, return to Stories, adjust decisions or notes, and generate again. Publishing is intentionally outside this dashboard and remains approval-gated.