ashvath

@diabhey you think I can get a copy of this 👀

It's 8am. Slack pings. My agent has read everything I clipped yesterday and sends me a digest of what's worth my attention. Later, when I open Claude Code to start a new project, it surfaces the saved articles that are actually relevant to what I'm building. Here's how I built "ab-brain".. The problem Like most people, I read a lot online. X, LinkedIn, Substack, tutorials, research threads. I bookmark them across phone, tablet and laptop, then never go back. Amazing articles from amazing people just rotting in bookmark folders. Felt like a shame not to fold that into my actual dev work and learning. So I built a pipeline. Capture: Obsidian Web Clipper on mobile turns any page into markdown and drops it into my @obsdmd vault. The vault lives in iCloud Drive, so I skip Obsidian Sync fees and get free cross-device replication. Sync: A launchd agent on my Mac (macOS's systemd) triggers rclone on an interval. rclone is rsync for cloud storage, keeping the iCloud vault in sync with a DigitalOcean VM. Provisioning: The @digitalocean droplet is spun up by Argus, the provisioning tool I've been posting about in the last few weeks. Classification: On the VM, a systemd file watcher picks up new notes, dissects and analyses them, runs searches to enrich sparse link-only notes, tags them, and indexes into @usemoss for semantic search. Every change is committed to a local git repo so I can revert anything. Runtime: Hermes agent runtime @NousResearch. Been a solid experience. Consumption: I DM the agent on Slack ("summarise Hermes agent notes that came in this week"). An 8am cron sends me a morning digest. On the Mac, a Claude Code skill pulls relevant saved notes into my coding sessions, so the agent I code with actually learns from everything I've been reading. Building similar workflows? Happy to chat. For deep engineering and technical storytelling, visit diabhey.com

@fullstackpython @usemoss @diabhey lets gooo!

Added a new pre-seed YC company, @usemoss to Plushcap: plushcap.com/moss. H/t to @diabhey for telling me about them. They’re working on pushing down the latency threshold on retrieval which is a blocker for conversational agents, especially Voice AI agents 🧵

@DrewVento the founding engineer boss

if that's what your founding engineers look like, you're gonna make it

@RuhanPonnada @ycombinator @oronila_ is that you??

I went to a conference with no booth and booked 6 next calls brought a putter and a hole: - if they won = free lulu merch - if they lost = tell me about their sales process as @ycombinator says, do things that don’t scale

@Phonenurd Apple’s new product- iVibe

Apple's at it again... 💀 ​I just downloaded the massive 18.4GB iOS 26.4 update on my iPhone 17 Pro Max, and it’s been stuck in a bootloop for the last 2 hours. ​All that for some new emojis and "stability"? Honestly, wtf Apple 😭 ​#iOS26 #Apple #iPhone17ProMax

@usemoss we’re hosting a meetup at SF tomorrow evening. skipping the panels and the pitch deck- just drinks and good conversations with other builders. luma.com/ha2xhx18

joined @usemoss recently and we immediately went full send 😭 we open sourced moss yesterday coming into OSS with: • 1,000+ users • 80k+ downloads > swag for first 100 merged PRs > inviting contributors to our SF office 👀 come try / break it → github.com/usemoss/moss

@trq212 @rowans_planet would there eventually be a feature to talk handsfree? it would also be rlly cool to talk to Claude when planning

Voice mode is rolling out now in Claude Code. It’s live for ~5% of users today, and will be ramping through the coming weeks. You'll see a note on the welcome screen once you have access. /voice to toggle it on!

your claude code’s context could be 100x better and look like this

@nathan_covey about time…

Breaking: LinkedIn has now discovered Claude

@ionleu I gotchu: trynebula.ai

drop ur startup link

@emilyhanyf @modal yoo could we also get more creds pls 🙏🥺

new @modal stickers dropped!

@adxtyahq @phytalia

when your cofounder is begging you to ship the AI MVP but you’re rewriting the backend in Rust because Python might struggle at 10M concurrent users

@Amank1412 @phytalia 🥺

In your 20s you'll meet a friend who introduces you to > APIs, > webhooks, > n8n, > Zapier, > AI/ML, > LLMs, > automation Make them your co-founder.

@nodvex backend is so much more easier, the solutions are more intuitive. Frontend is just magic lol if it works it works

Why most of the people choose frontend??

@adisingh hey! I’ll be in sf today. building memory for long horizon agents :)

It’s a beautiful, sunny day in San Francisco. I’m looking to get out of the house. If you’re a founder unhappy with or looking into anything email, I’m dedicating the day to visit you. Just reply and I’ll show up with a gift card + swag and come help!

@adxtyahq dude me too, checkout aldente to manage your battery life

got this mac last year already at 92% battery health am I cooked or is this normal

@_nicolealonso no way that’s literally my dock right now 😭

tech bro starter pack

@FurqanR what if Nebula uses Nebula for context management? trynebula.ai

This is why I’m working on Nebula. You don’t need to spend time on configuring and maintaining the system. Tell your agents what you need to do, everything else should manage itself.

@kakatohesss trynebula.ai

what are you working on? Drop your product below 👇 I'll repost a few and help you get in front of more eyes 🚀