Tell us what you care about — in plain English. Our agents go out overnight, scan hundreds of sources, cut the noise, and file your personal brief before you wake up.
So your morning takes six minutes, not sixty.
The Crawler
Scouts 50+ sources every night — news sites, Reddit, ArXiv, blogs, and more. It doesn't have a fixed list. It reasons from your interests and decides where to look.
The Filter
Reads everything the Crawler finds. Keeps what's relevant to you. Discards the rest. Out of 300 articles, you'll see 10. The ones that actually matter.
The Analyst
Doesn't just summarize. For every item it keeps, it writes one line explaining why it matters specifically to you — not to everyone, to you.
The Courier
Assembles your brief and delivers it. To your inbox, to your dashboard, at the exact time you choose. Every morning, without fail.
No invented headlines, no hallucinated citations. Every item in your brief links back to where it was actually published.
I care about
▌
Describe your interests
Open the Configuration tab and type what you follow. Plain English — no dropdowns, no categories.
Your agents deploy
Hit Deploy and watch them work on the Operations page — sources mapped, filters calibrated.
Your brief arrives
It lands on Operations every morning (and in your inbox, if you want). Past briefs live in the Archive tab.
It gets smarter.
The more you read, the better it knows you. Click what interests you, skip what doesn't — your agents adjust.
It respects your time.
Six minutes. That's the average time to read a full brief. Not an hour of scrolling. Not a 47-tab rabbit hole.
It's actually personal.
Not personalized by algorithm. Personalized because four agents are reasoning about your specific interests — not clustering you into a demographic.
It works while you sleep.
No prompting, no refreshing, no asking. Your agents run on a schedule. Your brief is waiting when you open your eyes.
This is what a brief looks like on arrival. Field reports, analyst notes, and nothing you didn't ask for.
Researchers found that models sometimes reason correctly in scratchpads while producing outputs inconsistent with that reasoning — raising questions about whether visible reasoning is the actual decision process.
Directly relevant to your interest in AI safety tooling — the gap between stated and actual reasoning is the core problem you have been tracking.
The new in-process memory layer lets long-running agents maintain state across interruptions without Redis or a vector DB. Early benchmarks show 40% reduction in round-trip latency for multi-step pipelines.
This lands exactly on the architecture problem you described — worth reading the migration guide before your next sprint.
The proportion of agent-infrastructure companies in the batch has doubled since S25. Most are targeting the orchestration and observability layer rather than foundation models.
The segment you follow is now 7.5% of YC's intake — the market signal you were watching for is here.
“Rust programming and generative music”
“Climate policy and venture capital”
“Retro game dev and Japanese cinema”
Tell us what matters to you. We'll handle everything else.
Deploy your agentsFree to start. No credit card required. Your brief arrives tomorrow morning.
Agents operate continuously. Ten items. Every morning.
MakeDigest