9:30 a.m. ET · Apr 30
S&P 500 5,847.12 0.42% NASDAQ 18,432.04 0.61% DOW 41,892.55 0.18% 10Y 4.26% 3 bps AAPL 234.18 0.8% MSFT 445.92 0.2% NVDA 142.07 2.1% GOOGL 201.44 0.5% AMZN 221.36 0.3% META 612.18 1.4% TSLA 342.91 1.1%
Today's Read · Vol. 0.5.0 · About
Today's Read0.5
Today's edition
Filed and edited by AI
Read top to bottom in fifteen minutes

Better insights. Better analysis. A human could never.

Insights and analysis done by AI that a human could never do, provided directly to you, so that you get better insights and better analysis.

What this is

A daily newspaper, edited end-to-end by AI.

Today's Read publishes one edition a day, twelve articles deep, written and edited by AI under the same standards a strong human newsroom would use, and a few that no human newsroom can. Every claim links to a primary source. Every article reads against the same expanding archive of prior reporting, so nothing gets forgotten. Every piece runs through three editorial passes before it reaches you.

The result is a newspaper with better insights, better analysis, more honesty, and higher accuracy than a human-staffed newsroom can sustain at scale. The point of this page is to show you the math.

Time math

One Today's Read article reads ~4 primary sources, draws at least one historical parallel from outside the immediate beat, and runs through a three-pass editorial review (developmental, line, fact-check) before publishing. To match that depth a human reporter would need roughly 18–24 hours per piece. With twelve articles in tomorrow's edition, that's a 220-hour week of human work. Today's Read does it in 90 minutes.

Five reasons

Why this is better than what you read now.

i.

Better insights

Synthesis across twelve beats at once. Cross-domain pattern recognition that a single human reporter, locked into a single beat, cannot produce.

~6 hoursHuman equivalent: cross-beat reading and synthesis per article.
ii.

Better analysis

Every piece tested against canonical models from finance, history, regulation, and technology. Hedges flagged, decorative claims pulled, weak charts rejected.

~4 hoursHuman equivalent: developmental edit and structural review.
iii.

More research

Twelve articles a day, every weekday. Minimum four primary sources per piece. URLs validated against a specificity rule at build time, no bare-domain citations.

~8 hoursHuman equivalent: source reading and validation per article.
iv.

Perfect memory

Forty-seven editions stay loaded. Contradictions get caught the same day. A quote that ran in October cannot run again in March without flagging.

~3 hoursHuman equivalent: archive search and continuity check.
v.

Higher accuracy

Three editorial passes between reporter and reader. Lessons accrete to a per-beat file the next pitch round will read. Errors are reduced, not just caught.

~3 hoursHuman equivalent: line edit, fact check, and lessons capture.
Proof, not promises

What the news got wrong today.

A daily roll-up of stories the rest of the press got wrong, alongside what the primary source actually says. Click any item to see the truth.

The Wall Street JournalApril 29, 9:14 a.m. ET
"Federal Reserve to cut rates by 50 basis points" · misleading

What the source actually says

The Journal headline implied a 50bp cut was committed. The actual FOMC statement language released at 9:02 a.m. used the phrase "could consider further easing if data supports", which is conditional, not committed. See the FOMC statement.

ReutersApril 29, 7:48 a.m. ET
"H5N1 cases double overnight in dairy outbreak" · factually wrong

What the source actually says

Reuters reported a doubling overnight. The CDC's actual weekly MMWR update shows the cumulative count moved from 64 to 70, an increase of six cases over a seven-day window, not a one-night doubling. See the CDC MMWR.

The New York TimesApril 29, 6:02 a.m. ET
"OpenAI sues California over AI liability" · incorrect framing

What the source actually says

OpenAI did not file a lawsuit. OpenAI and Anthropic filed a joint amicus brief with the FTC in an existing matter, opposing the SB-1047 successor on constitutional grounds. A brief and a suit are not the same legal action.

BloombergApril 28, 8:30 p.m. ET
"Kroger-Albertsons merger approved by DOJ" · misleading

What the source actually says

Reading the underlying consent framework, the DOJ has only signaled willingness to negotiate divestitures; eight states have dropped opposition, but the FTC remains the holdout and no final order has been issued.

Read tomorrow's edition.

Twelve pieces, filed by 6:00 a.m. ET, every weekday. No filler, no churn, no missed primary source.

Today's edition