Built this over the weekend because I subscribe to too many newsletters (Lenny's, Tomasz Tunguz, Data Driven VC) but never read them.
It's a Python/FastAPI backend that accepts forwarded emails, processes them with Gemini 3 (using its large context window), and stores results in Supabase to generate a structured Sunday briefing.
Right now it's an MVP. I'm looking for feedback on the summarization quality vs the original emails.
As someone who reads newsletters for fun, my suggestion would be to subscribe to less newsletters, and focus on the ones you enjoy reading the most. The product design seems to me motivated by FOMO and wanting to get all the info from all the newsletters. Think of newsletters less as a bucket you fill up and need to empty, and more as a river you can dip in and out of. Don't stress out about staying on top of all the hype. (I know you asked for product feedback but you got philosophical feedback instead.)
I love the 'river vs bucket' metaphor. You hit the nail on the head - FOMO was 100% the motivation here.
Ideally, I would just unsubscribe, but in my field (and as a student), I often fear missing a specific signal in the noise. My goal with this agent is actually to help me treat it more like a river: if the AI tells me nothing critical happened, I can skip reading without anxiety. Thanks for the perspective!
Nice! Just took a look. It seems we are attacking the same beast (newsletter fatigue) but maybe with different flavors.
I went down the route of using Gemini 3 specifically to generate 'strategic briefs' rather than just a feed. Curious what stack/models you're running for Hidden Signal?
It’s Django with a similar custom email that you can forward emails to or sign up through that email. And then uses open AI mini for summarization and nano for categorization. Then it sends you a digest with the summaries on your schedule.
Hey HN,
Built this over the weekend because I subscribe to too many newsletters (Lenny's, Tomasz Tunguz, Data Driven VC) but never read them.
It's a Python/FastAPI backend that accepts forwarded emails, processes them with Gemini 3 (using its large context window), and stores results in Supabase to generate a structured Sunday briefing.
Right now it's an MVP. I'm looking for feedback on the summarization quality vs the original emails.
You can try it here: https://sunday-dashboard.streamlit.app/ (It generates a unique email address for you to forward newsletters to).
As someone who reads newsletters for fun, my suggestion would be to subscribe to less newsletters, and focus on the ones you enjoy reading the most. The product design seems to me motivated by FOMO and wanting to get all the info from all the newsletters. Think of newsletters less as a bucket you fill up and need to empty, and more as a river you can dip in and out of. Don't stress out about staying on top of all the hype. (I know you asked for product feedback but you got philosophical feedback instead.)
I love the 'river vs bucket' metaphor. You hit the nail on the head - FOMO was 100% the motivation here. Ideally, I would just unsubscribe, but in my field (and as a student), I often fear missing a specific signal in the noise. My goal with this agent is actually to help me treat it more like a river: if the AI tells me nothing critical happened, I can skip reading without anxiety. Thanks for the perspective!
I built something similar! https://hiddensignal.app
Nice! Just took a look. It seems we are attacking the same beast (newsletter fatigue) but maybe with different flavors.
I went down the route of using Gemini 3 specifically to generate 'strategic briefs' rather than just a feed. Curious what stack/models you're running for Hidden Signal?
It’s Django with a similar custom email that you can forward emails to or sign up through that email. And then uses open AI mini for summarization and nano for categorization. Then it sends you a digest with the summaries on your schedule.
Nice work!