Are your competitors
already winning on Substack?

I help brands establish credibility on Substack through data-driven content strategies and original sponsored journalism.

Karen Spinner

Writer and content strategist with technical skills

I'm Karen Spinner, founder of StackDigest. I'm a writer and content strategist who learned about systems design, machine learning, and data engineering in the process of building my own AI-powered tools. May latest project is StackDigest, a Django-based web application with a vector database, semantic search, and ML-powered theme discovery across 2,000+ Substack newsletters and 39,000 articles.

Through building StackDigest, I gained hands-on experience with natural language processing, clustering algorithms, and vector similarity search (K-means, cosine similarity, pgvector). I also write about AI on Substack, documenting my experiments and what I'm learning about technical implementations.

My background is in content strategy and research. I've helped companies like Adobe, Confluent, and IBM develop content strategies and conduct market research. I've also grown a Substack newsletter from zero, with no existing email list, to 600+ subscribers in less than three months.

The services below combine what I've learned building technical tools with practical expertise in research, content strategy, and Substack growth.

Services

Substack intelligence report

Custom competitive analysis powered by StackDigest's backend infrastructure. Get deep insights into your competitors' content strategy on Substack.

  • Deep dive into competitor content and themes
  • Engagement metrics analysis (likes, comments, publishing frequency)
  • Machine learning-powered similar content discovery using cosine similarity
  • Quarterly updates to track changes over time

Custom research automation pipeline

Automated intelligence gathering on topics that matter to your business. I'll build a custom pipeline that monitors your industry and surfaces meaningful trends.

  • Pick any topic and your favorite sources
  • Vector database of daily news and articles, continuously updated
  • Machine learning to identify emerging themes
  • Weekly trend analysis and reports delivered to you

Custom research studies

Original research designed to answer your specific business questions. From data collection through final report, I handle the entire research process.

  • Custom data collection and analysis with machine learning
  • Detailed, well-researched reports with actionable insights
  • Product testing and experiment design
  • Transparent documentation of methodology and findings

Substack marketing assistance

Substack is a green field for influencer marketing that isn't yet overrun with brands. I can help you navigate this emerging channel with data-driven influencer partnerships.

  • Introductions to relevant influencers in your space (focus on tech/B2B)
  • Newsletter engagement and trend data to identify best partnership fits
  • Strategy for authentic brand presence on Substack
  • Leverage StackDigest's database of 2,000+ newsletters for discovery

Substack newsletter support for brands

End-to-end support for brands building their own Substack presence. From strategy to execution to performance tracking.

  • Content calendar development and management
  • Writing and research support
  • Results tracking and performance analysis
  • Data-driven content strategies based on engagement patterns

My recent projects

39,308
Articles indexed using
OpenAI embeddings
2,510
Newsletters in the
StackDigest database
20
Thematic trends detected
using K-means clustering

Machine learning analysis of StackDigest's database

To power content discovery in StackDigest, I built a vector database of 39,308 articles from 2,510 Substack newsletters using OpenAI embeddings and PostgreSQL's pgvector extension. The system performs semantic search to understand meaning and context, not just keyword matching.

Using K-means clustering on the embeddings, I identified 20 major thematic trends across Substack plus engagement patterns, which together provide actionable intelligence about which content types perform best.

This project demonstrates the kind of custom machine learning pipeline I can build for your business—from data collection through insight generation. Read the full analysis →

Reddit sentiment analysis: educators and AI

I conducted original research analyzing 415 Reddit posts from 10 education subreddits to understand teachers' genuine attitudes toward AI in the classroom. This wasn't survey data—it was authentic conversations from practicing educators.

The analysis used natural language processing to identify emotionally significant keywords and sentiment patterns. Key finding: 76% of emotionally charged mentions reflected negative or uncertain attitudes toward AI adoption.

This type of audience research provides product teams with real user sentiment that informs positioning and messaging strategy. See the methodology →

Building (and breaking) AI agent workflows

I designed and built a 4-agent copywriting system using Python and the Claude API to test whether AI agents could collaborate on long-form content creation. Each agent had a specific role: researcher, writer, editor, and fact-checker.

The result? The agents created infinite review loops, went off-brief, and generally failed at collaboration. But documenting the failure generated significant reader engagement and provided valuable insights about current AI limitations.

Transparency about technical experiments resonates with audiences. This project shows how I approach R&D work and communicate findings. Read the experiment →

Let's talk about your project

Interested in working together? I'd love to hear about your needs and discuss how these services can help your business.

Get in touch

Or reach out via Substack DMs