WHO WE ARE AND HOW WE WORK
Technology, data and journalism that respect your time.
MileiHoy was born from a simple question: how do you follow Milei's Argentina without spending the day jumping between news sites, official releases, X threads and INDEC spreadsheets? The reforms, the economic debates and the political fights move faster than anyone can read.
Our answer was to build a newsroom that combines automated reading, Artificial Intelligence and editorial rules written in advance. Technology does the heavy lifting of reading, sorting and comparing; the rules decide what gets published and how it is told, so what reaches you is what matters, with context and from every perspective.
How we work, step by step
01
We read everything, all day
Nine national outlets with different editorial lines, more than 40 official bodies (the Presidency, ministries, the Central Bank, Congress and the provinces) and 33 X accounts from the government and the opposition. The front page is refreshed all day long, not once in the morning.
02
We select before we analyze
Not everything that gets published deserves your time. We filter first: only stories with a national angle and a link to the government get through, and when several outlets report the same thing, we group it instead of repeating it.
03
We measure the framing
Every published story comes with a neutral summary and an estimate of how much of its approach leans left, center or libertarian, with the evidence from the text and the loaded terms it uses. It is an AI estimate, not a verdict on the outlet.
04
We put the versions side by side
With “One story, three views” you compare how each outlet headlines the same news, and with the I | C | L selector you choose which side to read the front page from, or see everything to contrast.
05
We listen to what is really posted
Social Alert and the X Cards are built from real posts: what the government and the opposition publish, what the President shares and who replies to whom. Every claim is checked against the post before it is published.
06
We check against official data
In the Economic Radar every figure going around is compared with the INDEC, Treasury or Central Bank data, and the verdict (matches, overstates, understates or contradicts) is computed by code, not by opinion. Market data is always labelled as unofficial.
07
Verified Institutional Photography
News photos are official press images, matched by topic to the release of the body that took the action, credited and linked to the source (Argentine Law 11.723). When the body only published a graphic card, we use a free-license illustrative photo (Pixabay). When a public figure posted the photo on social media and an outlet reproduced it, we use it crediting the account and the outlet. We also use photos the figure posted on their own account, chosen by the newsroom, with a link to the post. And free-license photos from Wikimedia Commons, chosen by the newsroom, credited to the author with their license.
08
In your language, at your pace
Three editions (Argentina, US Edition and Edição Brasil), semantic search, an edition-date filter and newsletters: the free Essential Summary, Premium with the full analysis and Intelligence with the briefing before the market opens.
What we don't compromise on
AI is disclosed
Everything a model estimates (framings, summaries, social cards) is labelled as such. We never present an estimate as a fact.
Every fact has a source
We cite the outlet, the official body or the post each thing comes from, with its link. If something cannot be backed by a source, it is not published.
The same yardstick for everyone
Framing is measured the same way for every outlet, and Data beats narrative applies to the press and to the government alike. We talk about figures and texts, never about people.
Methodology and trust
Each story's framing (left, center, libertarian) is an AI estimate based on the evidence in the text, not a judgment of the outlet.
Radar data comes from official sources (INDEC, the Treasury, the Central Bank). Market data comes from unofficial, delayed sources, and we always say so.
In Data beats narrative, code computes the verdict by comparing the published figure with the official one: it is about the figure, never the outlet, and it applies to the government too.
A community of readers who think
We don't want to tell you what to think, but to give you the tools to decide for yourself: each outlet's version, the protagonists' own words and the official numbers, all in one place. If it helps you, join the newsletters; if you know a source we should be reading, tell us.
The institutional pages use free-license illustrative photos from Pexels, credited to their authors.