How We Built This List and Our Ranking Criteria
We subscribed to and read each newsletter on this list for at least three months before including it. Assessment was based on five criteria: signal density (how much actionable content per newsletter, not just volume), editorial quality (does the author have genuine expertise or are they summarising what they read elsewhere), practitioner relevance (does the content connect to building AI products, not just tracking the industry), publication consistency, and free vs paid value distinction (some newsletters reserve their best content for paid subscribers, which is legitimate but worth noting). We deliberately excluded newsletters that are primarily fundraising round announcements or news aggregation. Those have their place but do not help a CTO make better technical or product decisions. We also excluded newsletters from large organisations where the content is shaped primarily by marketing considerations rather than editorial ones. The best newsletters in this space are written by practitioners with opinions, not by content teams optimising for brand safety.
The Full Ranked List: Pros, Cons, and Best For
1. The Pragmatic Engineer (Gergely Orosz). The highest-quality engineering leadership newsletter available. Deep, well-researched articles on engineering culture, career development, and technical leadership. AI coverage is increasingly prominent. Paid tier worth the subscription. Best for: engineering leaders wanting substantive analysis of engineering culture and leadership. Limitation: not exclusively AI-focused. 2. Lenny's Newsletter. The best product management newsletter, with increasing AI product content. Lenny Rachitsky interviews practitioners and distills frameworks. Best for: product-minded CTOs and technical founders. Limitation: broad product scope, not AI-specialist. 3. Import AI (Jack Clark). One of the longest-running AI newsletters written by a genuine AI researcher. Covers new research papers with lay-accessible summaries and implications. Best for: engineering leaders who want to track AI research developments. Limitation: research-heavy, less product-focused. 4. The Batch (Andrew Ng, DeepLearning.AI). Weekly digest from Andrew Ng covering AI news, research, and applications. Clear, well-edited, authoritative. Best for: engineering leaders who want a balanced weekly overview. Limitation: occasionally aligned with DeepLearning.AI's own products and perspective. 5. AI Breakfast. Daily AI news digest with a product and business focus. Good signal-to-noise ratio for staying current without deep dives. Best for: CTOs who want a quick daily briefing. Limitation: less editorial depth than weekly publications. 6. Interconnects (Nathan Lambert). Deep analysis of LLM research and AI policy from an RLHF researcher. Excellent for understanding what is happening at the frontier of LLM training. Best for: engineering leaders building on top of the latest model capabilities. Limitation: technically dense; requires ML background to get full value. 7. Last Week in AI. Accessible weekly summary of AI developments. Good for staying oriented without going deep. Best for: engineering leaders who want a reliable orientation newsletter. Limitation: less opinionated than the best newsletters on this list. 8. The AI Edge (Microsoft). Microsoft's AI newsletter for practitioners. Well-written with genuine technical content. Reasonable editorial independence despite the corporate origin. Best for: teams building on Azure or Microsoft infrastructure. Limitation: occasionally Microsoft-centric. 9. TLDR AI. Daily newsletter with very concise summaries. High frequency, low depth. Best for: engineering leaders who want rapid awareness of what is happening. Limitation: depth is sacrificed for volume. 10. ML Engineering (Eugene Yan). Infrequent but high-quality essays on production ML systems from an Amazon engineer. Best for: engineering leaders specifically managing ML engineering teams. Limitation: infrequent publication makes it less useful for staying current. 11. The Algorithm (MIT Technology Review). Journalistic coverage of AI from a credible technology publication. Good for context and regulatory/policy coverage. Best for: CTOs who need to track AI regulation and societal impact alongside technical developments. Limitation: journalistic perspective rather than practitioner perspective. 12. One Useful Thing (Ethan Mollick, Wharton). Thoughtful essays on practical AI use for knowledge workers and organisations. Less technical, more strategic and organisational. Best for: CTOs advising their organisations on AI adoption rather than building AI products.
Comparison at a Glance
Newsletters in this space occupy three distinct positions. Research-tracking newsletters (Import AI, Interconnects, The Batch) help you track what is happening at the model frontier. For a CTO making decisions about which LLM to build on, understanding capability developments before they become widely available matters. Interconnects is the most technically rigorous of these three. Practitioner and leadership newsletters (The Pragmatic Engineer, Lenny's Newsletter) are less AI-specific but more useful for the management and product decisions that occupy most of a CTO's time. Gergely Orosz in particular consistently publishes content that is more useful than the typical conference talk or podcast. News digest newsletters (AI Breakfast, TLDR AI, Last Week in AI) are best consumed as a quick daily or weekly scan. They are awareness tools, not learning tools. A CTO who reads only digest newsletters will know what happened but may not understand the implications. A practical newsletter stack for a CTO might be: The Pragmatic Engineer (weekly, paid) for engineering leadership depth, Import AI or Interconnects (weekly) for AI research tracking, and either AI Breakfast or TLDR AI (daily) for rapid awareness. This covers leadership, technical frontier, and current events without excessive time investment. UK-specific note: most of these newsletters are US-written and reflect US regulatory and market contexts. For UK AI regulatory developments, the Alan Turing Institute's AI UK updates and the DSIT (Department for Science, Innovation and Technology) newsletter provide UK-specific coverage of AI policy, including ICO guidance on AI and GDPR, the UK AI Safety Institute's work, and UK national AI strategy developments.
How to Choose the Right Option for Your Situation
Your newsletter stack should match what you are trying to accomplish. Three roles produce three different optimal stacks. If you are a CTO responsible for building AI products, you need technical depth on model capabilities and production AI systems. Prioritise: Interconnects (for LLM capability tracking), Import AI (for broader research context), and ML Engineering (for production ML systems when it publishes). Add The Pragmatic Engineer for engineering leadership. If you are a CTO whose primary role is enabling AI adoption across a non-AI company (insurance, retail, professional services), you need a different stack. Prioritise: One Useful Thing (for organisational AI adoption), The Batch (for accessible model updates), and The Algorithm (for regulatory and societal context). AI regulation in the UK and EU is moving quickly and has direct operational implications for companies deploying AI at scale. If you are a technical founder who is also the CTO, balancing product and engineering, the Lenny's Newsletter and Pragmatic Engineer combination covers most of what you need for the non-technical-but-critical decisions. For the technical AI layer, pick one research newsletter that matches your comfort level with technical content. Paid newsletter value: The Pragmatic Engineer is the clearest case where the paid tier is worth the approximately GBP 10 per month subscription. The depth of analysis in paid posts is substantially better than the free tier. Lenny's Newsletter paid tier is similarly high quality. For the rest, free tiers are adequate.
Our Recommendation
For most CTOs and engineering leaders in 2025, the optimal newsletter stack is three publications: The Pragmatic Engineer for engineering leadership (paid), Import AI or Interconnects for AI research tracking (free), and one digest newsletter for rapid awareness (TLDR AI or AI Breakfast, both free). This covers approximately one to two hours of reading per week and provides genuine signal on engineering leadership decisions, AI capability developments, and current events. Anything beyond this stack returns diminishing signal per hour of reading time. For UK engineering leaders, supplement this with ICO guidance updates on AI and data protection, which are published irregularly but have direct operational relevance. UK GDPR interpretation of AI systems is evolving, and staying current on ICO guidance prevents compliance surprises. Knowing what is happening in AI is necessary but not sufficient. Knowing how to build AI products that work in production requires practitioner experience and a team with the right technical depth. SpeedMVPs brings that depth to AI MVP delivery for UK founders. Get a free consultation at speedmvps.co.uk