/
Published on 06.03.2026
TLDR: The AI Break newsletter walks through building a complete AI-powered PR system using structured prompts - covering story angle mining, press release writing, journalist targeting, and follow-up sequences. It claims to compress weeks of agency work into about 45 minutes, though the real question is whether the output quality holds up without human PR expertise guiding it.
Look, I have been in tech long enough to watch every professional service get the "replace it with AI" treatment. PR agencies are the latest target, and honestly, the case made here is more compelling than most. The newsletter lays out a systematic approach to doing your own PR using AI as your engine: start by mining your business for genuinely newsworthy angles, write press releases in AP style, build targeted journalist lists, craft personalized pitches, and maintain a 90-day outreach calendar. The structure is solid - it mirrors what good agencies actually do internally.
What I find genuinely useful here is the prompt engineering approach. Rather than giving you a single "write me a press release" prompt, it breaks the PR workflow into discrete, specialized steps. The first prompt acts as a senior PR strategist extracting 5-8 newsworthy angles and ranking them by timeliness, impact, novelty, and human interest. That ranking framework alone is worth the read - most founders pitch what excites them, not what excites journalists. The distinction between trend pieces, data stories, founder journeys, and contrarian takes gives you a vocabulary for thinking about your media strategy that many first-time founders simply do not have.
The press release prompt is particularly well-constructed. It explicitly bans hype words like "revolutionary" and "groundbreaking," enforces AP style, and demands that quotes sound like actual humans talking rather than corporate ventriloquism. The rule about reading your press release out loud and rewriting anything that sounds like homepage copy is advice I have given dozens of times. The journalist targeting prompt builds a tiered outreach strategy rather than the spray-and-pray approach that makes reporters hit the block button.
But here is what the author is carefully stepping around: the gap between a structurally correct press release and one that actually gets picked up is enormous, and it is filled with relationship capital that no prompt can generate. Knowing that a specific TechCrunch reporter hates cold pitches but responds to Twitter DMs, or that a Forbes contributor is working on a trend piece about your exact space - that is the invisible value a good PR agency provides. The system described here will produce competent output, but competent output in a journalist's inbox competes with hundreds of other competent pitches every single day. The newsletter also glosses over the fact that AI-generated content is increasingly recognizable to journalists who read thousands of pitches monthly. If your "personalized" pitch reads like every other AI-assisted pitch, you have just automated mediocrity at scale.
The 45-minute claim is also worth scrutinizing. Yes, you can generate all these assets in 45 minutes. But the real work - actually researching journalists, building genuine relationships, iterating on angles based on rejection feedback, timing your pitches to news cycles - that is the job. The prompts are the easy part. The execution, follow-through, and human judgment are what separate coverage from crickets.