Nutgrafe
The news, reduced to what matters.
300 followers
The news, reduced to what matters.
300 followers
A "nutgraf" is the paragraph in a news story that explains what it’s about and why it matters. Nutgrafe does that for every article. Our AI reads every article from the world’s top news sites, blogs, and publications and delivers the core of each story in summaries of 400 characters or less, all in a familiar feed (circa 2015). Expand any post for key points and context, or click through to the original source to read more.
Really love the idea and the clean, elegant UI. Personally, I'm just reflecting on where it fits into my own use cases - but it's a lovely concept. Congrats on the launch! 🙌
Nutgrafe
@kiyaaa_ Thanks, I appreciate that. Nutgrafe was built around how many of us used social feeds for news (in the past), and grounding it back to that was the intent. Hearing how people think about where it fits is really helpful.
Hi Patrick, congrats on the launch!
I was taking a look and noticed you’re using Supabase. Just a quick heads-up, it looks like there may be some exposed data and possibly a few tables that don’t have RLS enforced yet.
I actually launched a small tool today called Supaleak that helps detect exposed secrets and Supabase misconfigurations like this. Might be worth a quick check to be safe.
Nutgrafe
@bscript Thanks for flagging, Bour. I did a full review of RLS and access patterns and everything is now enforced as expected. Auth data is handled by Supabase, and all payment info is processed via Stripe, so nothing sensitive is ever stored directly in the app’s database - mostly summaries, source links, and topics data. I’ll take a look at your product as well!
Congrats on the launch, Patrick! I’ve been looking for something that captures that '2015 Twitter' feeling, where you actually learned things rather than just seeing hot takes. The character limit is a brave but smart constraint; it forces the AI to focus on the essence rather than just a dry summary. It’s a huge time saver compared to digging through RSS feeds.
Nutgrafe
@yibao Thank you!
Nutgrafe
Quick update since launch 👋 @Nutgrafe
As time has allowed, I’ve been shipping additions based directly on feedback:
• Nutgrafe Daily – a free daily briefing at 9 AM ET with the biggest stories, summarized
• Dark mode
• Flagging – readers can flag summaries that need more context, improving accuracy over time
• Feed filtering – view just followed topics, everything, unread, saved, or by publication
I also slashed the price from $59/year to $24.99/year to make it easier for more people to try Nutgrafe long term.
You can now also follow updates at @nutgrafe on X.
Appreciate all the early feedback. Still shipping, still listening.
The focus on linking back to original sources rather than replacing reporting is a thoughtful approach. I'm working on something similar for Japanese tech blogs, so this resonates. Are there plans to expand beyond English-language publications, or is the AI summarization pipeline language-specific?
Nutgrafe
@yamamoto7 Thanks, appreciate that. We’re not locked into English. Right now most of the sources skew that way while I get the core experience right, but I’ve already started adding and testing international publications. I added The Japan Times and also 朝日新聞 (Asahi Shimbun) to see how they work in the feed. That said, if you have Japanese tech blogs or outlets you think should be included, I’d love the suggestions and can add them.
Makers Page
Name made me grin. I miss 2015 Twitter too. Quick nutgraf then move on — that’s my speed for the morning scroll. Like that it points back to the reporting. Curious about paywalled pieces and niche blogs. Gonna try it this week.
Nutgrafe
@alexcloudstar Appreciate that, Alex. That “quick nutgraf and move on” feeling is exactly the goal. I debated the name for a while, so I’m glad it resonates. Would love to hear how it fits into your week once you try it.
SigniFi
Congrats on the launch! 🚀 The idea of AI-generated 'nutgrafs' is brilliant for efficiency. My only worry with AI summaries is losing nuance or hallucinations. Do you have a human-in-the-loop process for top stories, or is it fully automated with specific prompts to ensure accuracy?
Nutgrafe
@yoang_loo Thank you! I think about this a lot. There’s no human in the loop today, but summaries are tightly constrained around what happened and why it matters. If the model doesn’t have enough context, it won’t generate a summary at all. I’ve seen that happen on a few articles, which honestly surprised me in a good way. I also review and reprocess when things look off, and longer term I’d like human review for top stories as it grows.