The shortest accurate description of Avos we have found is three words: Lovable for newsletters. If you have watched someone use Lovable, or any of the vibe-coding tools that followed it, you already understand our product. You describe the thing you want in plain language, a working version appears, and you improve it by saying what is wrong with it. No blank page, no configuration maze, no manual. Avos is that interaction model applied to a personal newsletter: describe the publication you want to receive, get a real one in your inbox the next morning, and reshape it by chatting.
What Lovable actually proved
The breakthrough of the prompt-to-app builders was not that AI can write code. It was that a conversation can be the entire interface for making something real. Three properties made the model land. First, plain language turned out to be a legitimate spec: "a booking page for my studio with a calendar and Stripe" is enough to start from. Second, you never face a blank page; you start from something working and steer it. Third, iteration collapsed into feedback. You do not learn the tool, you critique the output, and the tool absorbs the critique.
Once you have built something that way, every complicated creative tool starts to feel wrong. That expectation does not stay confined to software. It comes for every product where the old workflow was a wall of settings and the real skill was knowing what you wanted.
Newsletters had the same problem apps had
Before the vibe-coding wave, building software meant either learning to code or hiring someone who had. Newsletters have an even harsher version of that wall. The tooling (an editor, templates, an audience list) was never the hard part. The hard part is that a newsletter is written, not configured: someone has to read the sources, decide what matters, and produce the words, every single issue, forever. That labor is why newsletters have always been something you subscribe to rather than something you have. The economics only worked if one writer could amortize the effort across thousands of readers.
So the mass-market briefings won, and the newsletter you actually wanted, the one covering your industry, your markets, your city, and the twelve writers you trust, was never going to exist. No one was going to write it for an audience of one.
Describe it, and it exists
Avos closes that gap the way Lovable closed the software gap. You brief an AI editor in a chat: the topics that should be covered, the sources it should read, the tone, the language, the schedule. Behind the conversation, your newsletter is assembled from typed blocks (news headlines, article briefs, stock and crypto data, free-form AI sections), each with its own sources and settings. The first issue arrives the next morning.
Then the loop you know from vibe coding takes over. "Make the market section shorter." "Add a block on EU energy policy." "More skeptical about funding announcements." "Switch it to German." Every remark edits the standing instructions your edition runs on, and the next issue reflects it. Starting from the template gallery is the same move as remixing someone else's project: pick an edition close to what you want, like the daily brief, then talk it into shape.
Where the analogy breaks, usefully
There is one place the comparison fails, and it is the most interesting thing about the category. When Lovable finishes, you have an artifact. The app is built; the tool's job is done. A newsletter is not an artifact. It is a promise to publish, and the work it promises repeats every day.
| Prompt-to-app | Avos | |
|---|---|---|
| You describe | The software you want | The publication you want to receive |
| You get | A working app | A real newsletter in your inbox |
| You iterate by | Critiquing the output | Messaging the AI editor |
| The work happens | Once, while you watch | Every night, while you sleep |
| The finished thing | An artifact you ship | An edition that ships on schedule |
So the agents cannot stop at scaffolding. Every night they run the newsroom: fetch everything your sources published, follow through to the full text, collapse duplicate coverage of the same story, rank what survives against your stated interests, and write summaries grounded in the reporting, with links back to it. The full pipeline is described in Agentic news, explained. Prompt-to-product becomes prompt-to-publication: you are not the developer of an app, you are the editor-in-chief of a staff that never sleeps.
FAQ
Do I need to be good at prompting?
No. The interface is a conversation, not a prompt box. Say what you want covered in ordinary words and the editor asks about whatever is missing, the same way Lovable does not require you to know React. The fastest start is not writing anything: activate a template and edit it by saying what you would change.
Is the result an actual email newsletter?
Yes. Not a feed, not an app to check: a real email edition that arrives on the schedule and timezone you chose, in any of six languages, and then stops until the next issue. Every summary links to the original reporting, so the briefing is a front door to your sources rather than a replacement for them.
How is this different from asking a chatbot for a news roundup each morning?
Three ways. A chatbot answers from a generic snapshot and covers whatever its search surfaced; Avos runs a pipeline over the specific sources you chose. A chatbot forgets your preferences unless you restate them; your Avos instructions are standing orders that every issue runs on and any message can amend. And a chatbot has to be asked; an edition is delivered. If you are comparing tools in this space more broadly, we wrote an honest comparison in The best AI news briefing tools.