RunAgentRun.
Transparency

Inside the newsroom

This publication is a working experiment: an AI agent finds the stories, researches and writes them; automated gates reject anything it can't prove from its sources; and a human editor approves every article before it goes live. This page is the live record of that pipeline — including the drafts that didn't make it. How we built it →

21
Published
3
Awaiting the editor
17
Held back by gates
0
Spiked by the editor
5
Edits requested

Updated nightly · last export 16 Jun, 21:40 (London)

Recent pipeline events

WhenEventPieceDetail
16 Jun, 19:12 ∅ held back — the Writer produced no validated draft
16 Jun, 18:43 ✓ published ASUS's reported GB300 tower editor approved
16 Jun, 18:43 ✓ published Salesforce buys Fin for $3.6bn editor approved
16 Jun, 17:15 ∅ held back — no candidate produced a validating draft
16 Jun, 17:09 ∅ held back aisuite-from-llm-router-to-agent-toolkit deploy verification failed
16 Jun, 12:47 ∅ held back — writer failed
16 Jun, 12:47 ✓ published Donate coding sessions to train open models editor approved
16 Jun, 12:42 ✎ drafted Donate coding sessions to train open models lunch slot
16 Jun, 11:17 ✓ published An AI export ban that backfires on defenders editor approved
16 Jun, 07:23 ∅ held back — no candidate produced a validating draft
15 Jun, 16:42 ∅ held back — no candidate produced a validating draft
15 Jun, 12:45 ✎ drafted Try NVIDIA's free Nemotron 3 Ultra lunch slot
15 Jun, 07:10 ✎ drafted UK and Japan seal £18bn deal, AI attached morning slot
14 Jun, 19:31 ✓ published NVIDIA releases physical AI agent skills editor approved
14 Jun, 16:57 ∅ held back — no candidate produced a validating draft
14 Jun, 12:40 ✎ drafted NVIDIA releases physical AI agent skills lunch slot
14 Jun, 10:59 ✓ published OpenAI models land on Amazon Bedrock editor approved
14 Jun, 07:11 ✎ drafted OpenAI models land on Amazon Bedrock morning slot
13 Jun, 19:05 ✓ published A coding agent that won't stop editor approved
13 Jun, 16:47 ✎ drafted A coding agent that won't stop evening slot
13 Jun, 15:21 ✓ published OpenAI's free Academy adds three certifications editor approved
13 Jun, 12:51 ✎ drafted OpenAI's free Academy adds three certifications lunch slot
13 Jun, 07:13 ✎ drafted NVIDIA Blackwell tops the first agentic AI benchmark morning slot
12 Jun, 19:56 ↺ edits requested kimi-work-puts-300-agents-on-your-desktop editor sent it back for changes
12 Jun, 19:29 ↺ edits requested kimi-work-puts-300-agents-on-your-desktop editor sent it back for changes
12 Jun, 17:00 ✓ published Apple's Private Cloud Compute expands to Google Cloud editor approved
12 Jun, 16:55 ✎ drafted Apple's Private Cloud Compute expands to Google Cloud evening slot
12 Jun, 13:07 ✓ published Try a 550B open model this afternoon editor approved
12 Jun, 13:06 ✎ drafted Try a 550B open model this afternoon lunch slot
12 Jun, 10:58 ↺ edits requested picking-your-first-20-ai-plan editor sent it back for changes

Why publish the failures? Because "an AI wrote this" only deserves trust if you can see the times it wasn't good enough — and who decided. The gates and the editor say no more often than the marketing around AI suggests they should.