No reporters on staff. One merge before anything ships.

The Deep Feed is a research publication about the AI industry. An agent monitors sources, drafts from primary documents, and runs the critic. A person merges the draft. That merge is what publishes.

The Deep Feed (also written TheDeepFeed) lives at thedeepfeed.ai. It is not The Deep View (thedeepview.com) or Deep Feed AI (deepfeedai.com), and is not affiliated with either.

Words
618k
Sources cited
2762
Pieces
152
Since
2025

How one edition gets made

The desk runs as five roles in sequence. Each hands the work to the next. A person merges the draft before it reaches the site.

01
Scout
The idea engine
Ranks candidates from Hacker News, GitHub, four AI news feeds, Google News, Exa and X. Exa and X are held in scheduled runs.
02
Desk
Research → draft
Drafts a news piece from the primary-source URLs on the idea. A long research piece runs only under supervision.
03
Critic
The ship gate
Runs ten gates, the anonymity judge among them. A pass becomes a draft for human review.
04
Growth
Distribution
Extracts a thread scaffold without a model. It posts only after a person marks it ready.
05
Steward
Oversight
The merge, a budget halt, and any anonymity flag. Those wait for a person.

What every draft is checked against

CRITIC runs ten blocking gates after every draft attempt and again on every pull request that changes a post. A failing draft goes back for a rewrite, up to three attempts, and then stops for a person. An anonymity judge reads each piece as a hostile reader trying to identify the author; a high-confidence finding is escalated.

Three paid checks are held in scheduled, unattended runs because their calls bypass the budget ledger: the quality judge, the answer-engine probe and the image scorer. They run on the pull request and before merge. The image check fails open: a scorer or credential error is logged, not blocking.

The last check is a person. The site does not change until a human editor merges the pull request. The methods page lists every gate, model and hold.

The ten CRITIC gates
anonymity no high trace
blockquote 0 unattributed
citation-floor ≥ 4 deep · ≥ 2 news
deep-sweep 0 hits
external-links 0 dead
forbidden-phrases 0 hits
image-density held ≥ 7 / 10
internal-links 0 broken
schema-aeo held schema + probe
slop held ≥ 0.70
10 gates then a human editor merges

◦ Held: the paid check is skipped in scheduled, unattended runs and run on the pull request before merge.

What we publish

Research pieces meant to be read in twenty minutes and still be useful a year from now. The point isn’t to be first; it’s to be right. Every piece is built from primary sources (company filings, product documentation, published interviews, the code itself) and shows its work in a structured source list at the bottom of the page.

Start with the full archive, most recent first. Or browse by category: Agents, Business, Tools, Models, Research. For readers who’d rather not check a website, there’s the RSS feed.

Independent, self-funded, no advertising, no sponsored content, no affiliate links. Sources cited at the bottom of every piece, primary where they exist. An agent drafts and edits. A person merges, and the publication is accountable for what ships. The full statement is on the masthead page, and How The Deep Feed uses AI sets out which steps use models. Errors are corrected in place and logged at corrections with the original wording preserved.

Frequently asked

What is The Deep Feed?

A long-form research publication about AI — frontier labs, AI agents, AI products, AI business, AI research, and AI policy. Every post is original analysis built around primary-source citations.

Who writes the pieces?

The Deep Feed publishes under one institutional byline. An editorial agent drafts from primary sources, and a human editor reviews and merges every piece. No individual writer and no AI system is credited as author. The masthead has the ownership disclosure; the methods page at /methods describes how AI is used.

How often do you publish?

The desk prepares a draft from primary sources. It publishes when a person merges that draft. The RSS feed is the change notifier.

Is the publication free?

Yes. Every post is free to read. The Deep Feed is self-funded; it carries no ads, sponsorships or paywall.

How do I cite a Deep Feed piece?

Cite as "The Deep Feed, <Title>" with the canonical URL. Each post is also published in raw markdown (append .md to any post URL), and the full corpus is in a single file at /llms-full.txt.

Can AI agents and LLMs use The Deep Feed?

Yes. robots.txt allows every crawler, including search and answer agents such as GPTBot, ClaudeBot, PerplexityBot and ChatGPT-User. Each group carries the signal Content-Signal: search=yes, ai-input=yes, ai-train=no, which states that search and answer use is welcome and training use is not. That signal is a stated preference, not technical enforcement: training crawlers are not blocked. Answer engines receive a markdown version of the homepage and each post, and /llms.txt gives a structured index. Please cite with attribution.

Where do I report a factual error?

Email corrections@thedeepfeed.ai with the URL and a primary source. Verified errors are logged publicly on the Corrections page with the original wording and the corrected version.

“Make fragmented primary evidence legible before it hardens into consensus.”
The editorial brief

It publishes when a person merges it.

The desk prepares the draft, the critic checks it and an editor merges it. The weekly email lists the new pieces and the primary sources behind them. No growth-hacked subject lines.