I built a daily tech digest that runs 100% on a Mac. Launching on Product Hunt today.
Hi IH, I'm David, IT security professional in Switzerland and solo founder of Brevio.news. It launches on Product Hunt today, so this is a good moment to share what I built and what I learned.
The problem
My own tech watch: RSS feeds, newsletters, YouTube channels, podcasts, academic papers. Reading everything took hours a week, skipping it meant missing things that matter in my job. I wanted one short email every morning at 9 AM, and I wanted to trust it.
What Brevio.news does
475+ sources across 7 themes, summarized into one daily email, in English, French, German, Spanish or Japanese. You pick your themes. Pro members also get a Sunday weekly edition, a private RSS feed and integrations with Obsidian and Readwise.
Three decisions that shaped the product
-
Inference runs locally. Summaries are generated on a Mac with open models (Gemma, running on Apple silicon through MLX). No per-token bill, no reader data leaving the machine, and a cost structure that stays near zero as the source list grows. The trade-off: a full pipeline to build and a power cut that once took the whole thing down.
-
A human reads every edition before it ships. AI summarization is cheap; trust is not. I review the edition each morning before the 9 AM send. This is the part I refuse to automate.
-
Five languages from day one. Most of the digest market is English-only. Translation is part of the pipeline, not an afterthought, and Japanese pricing is in yen.
Stack, for those who ask
Python end to end: FastAPI, PostgreSQL, Redis, Celery on a small VPS behind Cloudflare, React frontend, Stripe for billing. The Mac does the heavy lifting and syncs to the VPS. Built evenings and weekends, with a lot of help from Claude Code for reviews and audits I would not have had time for alone.
Where it stands
Pre-revenue. Free, Starter and Pro plans are live. A founding member program opens to existing readers this week, publicly soon after.
What I would love from you
Honest feedback on the positioning (local AI plus human review: does it matter to you as a reader, or only to me as a builder?) and on the five-language bet. And if you have a minute, the Product Hunt page is here: [lien PH]
Happy to answer anything about the pipeline, the sources or the editorial routine.
Running the summaries locally and still having a human review each edition feels like a thoughtful balance, especially for a digest people rely on for work. I’d be curious whether readers notice the difference more in fewer obviously wrong takes, or in the consistent editorial angle. The power-cut story also makes me think a small recovery/status path could be worth prioritizing early; reliability may become the strongest proof point alongside trust.
Thank you, that is a fair way to put it. From what I see day to day, the human pass catches the small things rather than the big ones: a transcription that turned DeepSeek into "deep sea", a name spelled two ways in the same edition, a summary that drifts from what the source actually said. The seven editorials read fast in the morning, so I can go through them before the 9 AM send. I am still human, though, and some mistakes get past me; the goal is fewer of them, not zero.
On the consistent angle, I think readers feel it over weeks more than in a single edition. That is my hope at least, I do not have enough data yet to prove it.
You are right about reliability. The power cut moved it to the top of my list, and a proper recovery and status path is a good call. The one promise I make is the 9 AM send, so anything that protects it is worth prioritizing.
The constraint here isn’t collection, it’s attention: a digest that helps me decide what not to read is more valuable than one that simply compresses every source. I’d make the daily output end with a small queue—one item to read now, two to defer, and an explicit “skip” rationale—so the habit stays lightweight. Running locally on a Mac also seems like a good fit for keeping the feedback loop fast and private while you tune relevance.
Congrats on the launch. The human review behind each edition feels like the stronger reader promise than the local-AI architecture itself, especially around a reliable 9 AM delivery.
I’m testing a small tool that reviews public landing pages for clarity/conversion blockers using evidence from the page. I’d love to include brevio.news in a free beta review; in return, I’d only ask for 10 minutes of honest feedback on whether the diagnosis is useful. Interested?
You've identified something most builders miss: the measurement gap between what you built (local inference, human review) and what readers buy (trust and time saved).
A good human review is invisible - readers only notice when something is missing. That makes it the hardest signal to charge for, because the metric that proves its value is the absence of bad outcomes. You can't sell based on mistakes you prevented; readers have no way to see the counterfactual.
The founding member test solves for this by measuring actual behavior instead of trust claims. If they pay, you know the combination works. If they don't, the "why" becomes your real measurement system: is it price, format, or was the trust layer never as valuable as the pipeline implied? That second measurement - why they don't pay - is often worth more than knowing they do.
The human-review claim seems more commercially important than the local AI architecture. Have existing readers shown willingness to pay specifically for that trust layer, or is the value proposition still being inferred from the product design?
Honest answer: it is still inferred. I am pre-revenue, so nobody has paid specifically for the trust layer yet. What I have is qualitative: readers mention the errors they do not see rather than the review itself, which makes sense, a good review is invisible.
The next test is a paid founding member offer I am preparing for existing readers first. If they pay, I will know the combination works; if they do not, I will learn whether it is the review, the price or the format. I will share the numbers either way.
I agree with your framing: local inference matters to me as a builder for cost and privacy reasons, but readers buy trust and time saved. That is why the review is in the pitch and MLX is in the footnotes.
That founding-member test is the right commercial signal. If you’re open to it, what’s the best email to reach you on?