LinkedIn Content Engine: Research-to-Draft Automation
How an autonomous content engine sustains 2-3 quality-gated posts a week on under 20 minutes of founder time
Executive Summary
Sustained quality-gated posting cadence, up from near zero
Founder approval time per week
Validated ideas captured against defined content pillars
Minimum rubric score (of 100) for every published post
The Challenge
The founder of a B2B technology consultancy needed a consistent LinkedIn presence to build authority with founders, CTOs, and enterprise decision-makers — but content production competed directly with client delivery work. Before the system, posting had effectively stopped: research, drafting, redrafting, and scheduling could not compete with billable work. The account was in a drought, with impressions down 47% and profile views down 71% during the period before the system was built. The target was 10-11 posts per month, but posting became irregular and eventually near zero. Good ideas were lost because no system captured or prioritized them, and every post required starting from a blank page. The goal was not to remove the founder's voice — it was to remove the blank page, the scattered research, and the mechanical work around production so the founder could spend attention only on topic judgment and final approval.
Key Pain Points
- ✕Posting had effectively stopped — impressions down 47%, profile views down 71%
- ✕Content production competed directly with billable client delivery
- ✕Good ideas lost in scattered notes with no capture or prioritization system
- ✕Every post started from a blank page and gut-feel quality judgment
Our Solution
Amasa built a three-stage autonomous content engine covering research, ideation, prioritization, drafting, and quality gating, with the founder as sole approver. Every weekday morning, the collector scans Twitter, Hacker News, RSS feeds, and news sources, validating candidates against four content pillars: AI Product Strategy, Automation, Engineering, and Founder Insights. Five to ten ideas land in a Google Sheets Ideas Bucket and are summarized in Slack. An hour later, the topic extractor groups ideas by theme, scores and prioritizes them, and sends the top three topics to Slack — the founder approves topics in about five minutes. On Fridays, the auto-drafter writes posts for approved topics and scores each draft against a seven-criterion, 100-point rubric: hook strength, audience relevance, insight novelty, specificity and proof, comment potential, readability, and format fit. Posts scoring 85+ move to the Content Calendar; 70-84 triggers one automatic rewrite; below 70 goes to a Needs Review tab with concrete feedback. The system runs on Claude-powered skills orchestrated by TypeScript CLI tooling, with Google Sheets as the system of record across five operational tabs, fail-loud saves with full per-run transcripts, and randomized posting windows. As posts accumulate performance data, real engagement feeds back into topic and format scoring.
Implementation Approach
Editorial Strategy & Pillars
Defined the four content pillars and the five-tab Google Sheets system of record for ideas, calendar, digests, review, and performance
Research Collector
Built the weekday collector scanning Twitter, Hacker News, RSS, and news APIs with pillar validation and Slack digests
Topic Extraction & Approval Flow
Shipped theme grouping, scoring, and a five-minute founder approval loop over Slack and Sheets
Auto-Drafting & Quality Gate
Built the Friday drafter with a seven-criterion 100-point rubric routing drafts to ready, rewrite, or review
The Results
The implementation delivered transformative results across all key metrics, with immediate impact on operational efficiency, accuracy, and customer satisfaction.
Impact Metrics
| Metric | Before | After | Improvement |
|---|---|---|---|
| Posting consistency | Near zero and irregular | 2-3 quality-gated posts per week | Drought reversed |
| Founder time | Content deprioritized entirely | Under 20 minutes/week | Approval-only role |
| Quality control | Gut feel | 85+ score on a 100-point rubric | Objective gate |
| Cost of equivalent output | $1-2K/month agency or ghostwriter | Internal automation on API costs | Retainer avoided |
Key Takeaways
- Keep the founder's voice by automating everything except the two decisions that matter: which topics and which final drafts
- A 100-point rubric with an 85+ publish threshold replaces gut feel with an objective, explainable quality gate
- Daily pillar-validated idea capture builds a durable research backlog instead of losing ideas in browser tabs
- Engagement data feeding back into topic scoring makes the engine compound rather than just stay consistent
Inside the Engagement
The Workflow Before
- Scan Twitter, Hacker News, and industry news for angles.
- Capture ideas in scattered notes.
- Decide what was worth writing about.
- Draft posts from a blank page.
- Rewrite until the post sounded sharp.
- Judge quality by gut feel.
- Find time to post between client commitments.
The target was 10–11 posts per month, but posting became irregular and eventually near zero. Good ideas were lost because no system captured or prioritized them, and every post required starting over.
The Workflow After
- The weekday idea collector scans APIs and feeds.
- Ideas are validated against the defined pillar mix.
- Google Sheets stores the Ideas Bucket, Content Calendar, Digest Log, Needs Review, and Performance Tracking tabs.
- Slack sends daily digests and top topic candidates.
- The founder approves topics in about five minutes.
- The Friday auto-drafter generates posts for approved topics.
- The quality rubric routes drafts to ready, rewrite, or review.
- The founder performs final sign-off and schedules ready posts.
- Real engagement data feeds back into topic scoring once enough posts are tracked.
Full Results
| Metric | Before | After |
|---|---|---|
| Founder time | Content deprioritized; posting had stopped | Under 20 minutes/week for 2–3 posts |
| Posting consistency | Near zero and irregular | 2–3 quality-gated posts per week |
| Quality control | Gut feel | Every published post scores 85+ on a 100-point rubric |
| Research pipeline | None; ideas lost | 5–10 validated ideas captured daily |
| Cost of equivalent output | $1–2K/month agency or ghostwriter retainer | Internal automation running on API costs |
| Reach context | Impressions down 47%, profile views down 71% | Posting drought reversed with sustained cadence |
| Human role | Research, draft, rewrite, schedule | Approve topics and final posts |
The Impact
The founder’s scarcest resource is attention. The content engine redirects that attention to the two decisions that matter: which topics are worth publishing and which final drafts represent the brand.
The system also creates a durable research backlog – ideas are captured and scored every weekday against a defined editorial strategy instead of vanishing into browser tabs and notes. And the value compounds: as posts accumulate performance data, the engine replaces static scoring defaults with actual audience response by format and topic.
The Takeaway
Amasa turned founder-led LinkedIn marketing from an inconsistent blank-page task into an autonomous research-to-draft engine – sustaining 2–3 quality-gated posts per week with less than 20 minutes of founder approval time.
Quick Facts
Industry
Professional Services
Solution Type
Generative AI
Published
August 24, 2026
Technologies Used
Related Resources
Ready to Achieve Similar Results?
Let's discuss how we can transform your business with AI solutions tailored to your specific needs