AI Content Curation: Build a Self-Updating Knowledge Feed

Most knowledge workers spend 2–3 hours per week hunting for content.

Google for relevant articles. Browse Twitter/LinkedIn. Check newsletters. Ask colleagues.

This is inefficient.

Better approach: Build a system where content comes to you.

AI can curate a personalized knowledge feed that:

• Surfaces relevant content automatically
• Filters out noise
• Prioritizes by relevance
• Learns your preferences

This guide covers building an AI-powered content curation system.

---

What AI Content Curation Does

The Process

1. Define sources: Choose where to pull content from (RSS feeds, newsletters, APIs, databases)
2. Set criteria: Define what "relevant" means (keywords, topics, quality, recency)
3. AI ranks: AI reads new content, scores relevance, prioritizes
4. You review: Weekly, you scan top-ranked items
5. AI learns: From what you read/ignore, AI improves rankings

Example

Your criteria:

• Relevant to: AI, knowledge management, productivity
• Quality: Only established sources
• Recency: Published in last 2 weeks

AI does:

• Scans 500 articles/week from your sources
• Ranks all 500 by relevance (1–10 score)
• Returns top 10
• You read 2–3

Result: 10 curated items instead of manually hunting through 500

---

What AI Does in Curation

Task 1: Relevance Ranking

AI reads article headlines + summaries.

Scores: "How relevant is this to my interests?" (1–10)

Example:

Article: "How to Scale PostgreSQL on Kubernetes"

Your interests: Cloud infrastructure, databases

AI score: 8/10 (hits two interests)

Task 2: Deduplication

Multiple sources publish similar articles.

AI identifies: "This is the same story, just different source"

You see: One article, not five duplicates

Task 3: Clustering

Groups related articles:

• "AI Regulation News" (5 articles)
• "New LLM Benchmarks" (3 articles)
• "Developer Tools" (7 articles)

You quickly scan clusters. Read one representative from each.

Task 4: Summarization

For each article, AI generates 1-sentence summary.

You quickly assess: "Is this worth 5 minutes?"

Task 5: Prioritization

AI ranks not just by relevance, but by:

• Timeliness ("Is this trending today?")
• Authority ("Is this from a credible source?")
• Novelty ("Does this present new information?")

---

Building Your AI Curation System

Step 1: Define Your Content Interests

Write down:

• Topics you care about (e.g., AI, product management, leadership)
• Problems you're solving (e.g., "how to scale a remote team")
• Skills you're developing (e.g., "sales fundamentals")
• Competitive intelligence (e.g., "what's my competitor doing?")

Be specific. "Everything" doesn't work (too much noise).

Step 2: Identify Content Sources

Where does good content exist?

• RSS feeds (personal blogs, publications)
• APIs (Product Hunt, Hacker News, ArXiv)
• Newsletters (your subscriptions)
• Social (Twitter, LinkedIn)
• Databases (academic papers, industry reports)

Start with 5–10 sources. Add more later.

Step 3: Set Up Automated Pulling

Use a tool to automatically collect content from your sources:

Simple approach:

• Use IFTTT or Zapier
• Create rule: "Add RSS feed items to a Notion database"
• Articles auto-populate daily

More technical approach:

• Use an API aggregator (e.g., News API, Guardian API, Hacker News API)
• Collect articles into a database
• Run daily

No-code platform approach:

• Use specialized tools like Feedly or Inoreader
• Curated feeds do much of this already

Step 4: Configure AI Ranking

Set up AI ranking (depends on your tool):

If using Notion:

• Create a formula that scores relevance
• Input: keywords matching topics
• Score: articles with more keyword matches rank higher

If using specialized curation tool:

• Tool has built-in ranking
• Configure your interests
• AI learns over time

If using API approach:

• Use ChatGPT API to score articles
• Batch process new articles daily
• Return ranked list

Step 5: Create a Weekly Review Ritual

Friday Morning:

1. Open your curated feed (15 mins)
2. Scan top 20 items (2–3 mins each headline review)
3. Mark as "read," "for later," or "skip"
4. AI notes what you read

Over time, AI learns your preferences.

---

Avoiding Curation Failures

Failure 1: Filter Bubbles

AI only shows what matches your interests perfectly.

You miss novel ideas from adjacent fields.

Prevention: Include "serendipity slots"

• 80% curated by AI (high relevance)
• 20% random from your sources (low relevance, but diverse)

Failure 2: Overpersonalization

AI learns your preferences so well, you only see articles you already agree with.

No exposure to different perspectives.

Prevention: Set a rule: "Include at least one contrarian source"

• Subscribe to a publication with opposite politics/ideology
• Set rule: "Always include one from this source"

Failure 3: Quality Decline

Your source list decays. Blogs go inactive. Publications decline in quality.

Prevention: Audit sources quarterly

• Are they still publishing?
• Is quality still high?
• Remove underperformers
• Add new high-quality sources

Failure 4: Noise Accumulation

Source list grows to 50 sources. AI can't prioritize anymore.

Too much content to filter.

Prevention: Keep source list intentional

• Start with 5
• Add only sources you've manually verified
• Remove sources that don't earn their slot
• Maintain discipline on source count (target: 10–20)

---

Maintaining Feed Quality

Monthly: Review AI Rankings

1. Look at top 20 this month
2. Would you have picked these manually?
3. Any wrong rankings?
4. Adjust criteria if needed

Quarterly: Audit Sources

1. Are your sources still relevant?
2. Have any declined in quality?
3. Any gaps in coverage?
4. Add/remove sources accordingly

Quarterly: Review Interests

Your interests evolve:

• New project = new interests
• Skill mastery = shift focus
• Market changes = adjust priorities

Update your defined interests quarterly.

---

Tools for AI Curation

Option 1: Feedly or Inoreader (Simplest)

What it does: Aggregates RSS feeds, AI recommends

Pros: Easy, handles 90% of use case

Cons: Limited customization

Option 2: Notion + Zapier (Semi-Custom)

What it does: Automate articles to Notion, create ranking formula

Pros: You control the system

Cons: Requires configuration

Option 3: Custom API Stack (Most Control)

What it does: Pull from multiple sources via APIs, use ChatGPT to rank

Pros: Complete customization, fully AI-powered

Cons: Technical setup required

Option 4: Specialized Tools

Example: Perplexity Labs, Morning Brew (newsletter-style), Substack curation

Pros: Done for you

Cons: Limited to their coverage

---

The Workflow in Practice

Daily (5 minutes)

Check: Did new high-ranked articles come in? (usually, just notifications)

Weekly (20 minutes)

Friday morning:

1. Open curated feed
2. Scan headlines (10 mins)
3. Mark read/later/skip (5 mins)
4. Queue 1–2 articles for deep reading

Monthly (30 minutes)

Review AI rankings and adjust criteria

Quarterly (1 hour)

Audit sources and interests

---

Realistic Expectations

What AI Curation Does

✅ Reduces time spent hunting for content (from 2–3 hours to 30 mins/week)

✅ Improves content quality (you see curated, not random)

✅ Learns your preferences (better recommendations over time)

✅ Prevents important content from slipping through cracks

What AI Curation Doesn't Do

❌ Replace human judgment (you still decide what matters)

❌ Eliminate bad sources (quality of sources matters more than AI)

❌ Guarantee serendipity (needs deliberate inclusion of unexpected)

❌ Replace deep research (AI curation is discovery, not depth)

---

Starting Your AI Curation System

This Week

1. List 5 topics you care about
2. Identify 5 RSS feeds or sources for those topics
3. Set up Feedly or similar (free tier)
4. Subscribe to your sources

Week 2

1. Let system run
2. Review articles that come in
3. Assess: useful? Too noisy?
4. Adjust sources if needed

Week 3+

1. Establish Friday review ritual
2. Let AI learn your preferences
3. Monthly review and adjustment

---

Conclusion

AI content curation surfaces relevant content automatically, saving hours per week.

Build it:

1. Define your interests (5 topics)
2. Identify sources (5–10 RSS feeds, newsletters, APIs)
3. Set up automated pulling (Zapier, IFTTT, or specialized tool)
4. Configure AI ranking (tool-dependent or manual formula)
5. Weekly review ritual (Friday, 20 mins)

Maintain it:

• Monthly: Review AI rankings
• Quarterly: Audit sources and interests

Start this week:

1. List topics
2. Identify sources
3. Set up tool
4. Let it run for a week
5. Assess usefulness

In a month, you'll have a working curation system that saves 2+ hours/week.

For more on curation, see Content Curation Complete Guide. For information diet, check Information Diet Design.

Curate automatically. Prioritize intelligently. Learn faster.

Let relevant content find you.