Runtime Wire, which launched in May, is an artificial intelligence newsroom covering the AI economy, with Ryan Merket serving as founder and final editor.
It covers funding, model launches, developer tools, infrastructure, executive moves, court filings, and startup security. It also produces daily video and audio and sends reporting into developers’ code editors.
Since launching, Runtime Wire has published 1,627 articles and logged 205,000 article pageviews. It has scanned 71,796 potential stories and published just 2.3%.
“I’m the founder and the only human,” said Merket. “The newsroom itself is software — a pipeline of AI systems that scans sources, researches, writes, edits, fact-checks, and produces the video and audio around the clock. My job is part editor-in-chief, part engineer. I set the editorial standards, steer and review coverage daily, and build the systems that do the volume work.”
Runtime Wire has yet to produce revenue, but Merket says the monetization surfaces are built and live, including sponsored placements inside code editors (VS Code, Claude Code, terminal tools), a jobs board, and site sponsorships.
Merket is the former chief technology officer at Microsoft for Startups. He also worked at Reddit and Amazon Web Services and is a founder of Comal Ventures, an angel investor.
“I’ve been running the ad system largely on house inventory first to tune the mechanics — pacing, quality scoring, bot filtering — before selling it in earnest,” said Merket. “The plan is B2B-first: developer-tools advertisers, sponsors, and job listings. Reader access stays free, and we publish our operating metrics at runtimewire.com/in-public.”
The stories come from about 50 monitored sources — vendor blogs and changelogs, research feeds, social media and tips. Off-topic material (sports, crypto price talk) is rejected before scoring even starts.
Everything else is scored by an AI curator against a written editorial standard: is it new, does it matter to people who build with AI, is it from a live primary source, and has Runtime already covered the underlying event? Two separate layers catch re-reports of the same story. High-signal items from official vendor channels can take a breaking news “fast lane.”
“I also assign stories directly when I spot something the feeds miss,” said Merket.
Here’s how the process works:
The life of an article:
Selection — A story clears the curator, or Merket assigns it.
Research — The system fetches and reads the primary sources: the actual announcement, linked documents, and even attached images, which get transcribed so image-only claims can be verified. It builds a research brief anchored to today’s date, so old news can’t masquerade as breaking.
Draft — A writer model with live web search drafts the headline, summary, and body from that brief, steered by editorial guidelines: no hype, every claim sourced, U.S. audience-first.
Edit and fact-check — On the automated lane, an editor model reviews for depth and accuracy and can send the piece back for more research; a separate fact-check pass verifies specific claims against the live web. On stories Meket assigns and publishes directly, he’s the editor and reviews before it goes out.
Publish — The article goes live with a hero image, internal links, and structured data. Then the multi-format engine takes over: audio narration, translations, newsletter, podcast, the daily anchored video episode, Shorts and TikTok, SMS alerts for breaking items, and headlines distributed inside code editors.
