Ed Moss @mossifiedd

Claude outage for two hours, guess that means weekly usage limit reset.

@camjmann yeah the way I describe it, it is like living in a city that is a growth stage series C startup - especially there area I’m in (midtown/edgewater)

@SCR01111 Also feeny.ai

Just use better-auth, drizzle, postgres and deploy via docker, coolify & hetzner.

Caste @castedev

a week ago

I will never use Supabase again. It's free... until you have too much data. Then they lock EVERYTHING - even the delete button. Couldn't free my own space. Couldn't export my own files. One option left: pay $25. That's not a free tier. That's a hostage tier.

237 9 438 69K 83

If you don’t open-source and continue to guardrail, all IP and existing open source projects face threat. But as a result, you might see projects going closed-source in the future to protect against IP & security. Tricky dilemma to be in.

Chaofan Shou @shoucccc

2 weeks ago

Kimi K3 exploited the latest Redis server with a 0day it discovered. All it took was 27min with 32 agents. github.com/berabuddies/re…

109 454 5K 1.4M 2K

My mom still thinks I’m a “website designer”

@kaledora Good luck! Rooting for you all! 🙏

@AT_Awet No

I’m coding 12-16 hours a day, averaging about 50-100 commits, and thousands of lines of code. Nearly 7 days a week. But I can barely hit >50% of my weekly usage on a $200/mo Clade Code max plan. How are engineers @ startups spending so much money on LLMs? Is it all just API?

The end-state of most agentic platform interfaces will likely be just notifications. Let agents work, iterate and hit goals in the backgrounds, so you only need to be notified of it's progress - ideally via slack or imessage/whatsapp.

@justinstorre On the opposite trend, for those who use @WisprFlow - we should do a no-keyboard no-mouse hackathon.

Check porkbuns API reference porkbun.com/api/json/v3/do…

Hey @ctatedev @rauchg - would be really awesome if we could enable buying domain names via vercel.com/domains (my favorite domain search) with an agent

@tursodatabase @DrizzleORM @paradedb @lightpanda_io Forgot @paysponge = agentic payments w/ x402

I’ve been enjoying this tech stack lately for building agentic tools: - @tursodatabase & @DrizzleORM = SQLite db with generous free tier. No docker or VPS needed. Otherwise @paradedb if going with Postgres for a beefier setup. - agentbrowser & @lightpanda_io = super fast headless browser. - @better_auth = simple auth setup. - @supermemory = memory and context engine. - @vercel = for Eve (agent framework), AI SDK & Chat SDK. - @OpenRouter = access any model

And part two here: x.com/mossifiedd/sta…

Ed Moss @mossifiedd

4 weeks ago

Here is part two on GTM hacking. You can see my previous post (linked below) for part one. We last left off at a $1 budget with no pipeline, but we have a list of b2b companies that have been vetted & qualified through an LLM. 6) Now we need to get linkedin profiles of some

2 0 3 656 2

You can read part one here x.com/mossifiedd/sta…

Ed Moss @mossifiedd

4 weeks ago

PART ONE: Lately, I've been going deep on GTM engineering. Having done so much of it in the past, especially with a background in growth, product design & engineering, everything I've learned just pretty much "clicked". But the wildest part is how much money teams are

2 0 2 686 1

Time for part three on GTM hacking: We're finally at the stage where we have our ICP, our list of b2b companies, contacts and emails. You can read part one and two in the comments section below. So far we've spent roughly $3 & some change (variable LLM pricing). Give or take a few cents, let's say our budget is down to $1.50, this is still plenty to work with. The last step we left off on was personalized outreach. Here's how we achieve this. 9) In order to make anything feel truly personal, you need strong signals. So let's grab some. - First, we crawl your entire site, with adaptive LLM crawling (this means we don't explicitly define URLs & page paths, instead we give an LLM a goal and tell it to crawl up to a certain page depth until it feels satisfied that the goal was met). And you guessed it, we'll just use another github repo and Claude will continue to do all of the work for us. Here we'll use Crawl4ai: `github[dot]com/unclecode/crawl4ai` which also has an MCP and a CLI tool. This gives us a homegrown Firecrawl that we can run anytime without credits. - Then, we'll give it a website, ask an LLM to search for specific things, and turn all of it into markdown so it's readable. This all comes out of the box, for free (our favorite word). 10) Now before we move onto other signals, let's set up a foundation that makes our signal/data collection more useful. Every B2B "BOOK A DEMO" GTM SaaS tool is going to hate me for this. And FYI, if you've been following since part one - other than variable LLM pricing, we still haven't spent any unallocated budget since part one, so we can actually provide our LLM with more funds, an entire $1 worth! And to recap, we now we have a framework for: - Finding b2b leads (via job descriptions) and linkedin leads with our homegrown SerpAPI / Serper. - Finding & validating emails with proxies & our homegrown MillionVerifier. - Deep crawling websites with our homegrown Firecrawl. Now let's build our homegrown custom GPT. We achieve this with yet another open-source github repo called OpenRag: github[dot]com/langflow-ai/openrag/ This mainly gives us a full RAG pipeline system, with document ingestion and basic chat/query interface. We'll use a super cheap model like Qwen3 Embedding 8b ($0.01/m tokens): openrouter[dot]ai/qwen/qwen3-embedding-8b - to handle all of our embedding needs. Congratulations! Now we can feed our homegrown custom GPT any signal, and query against it in a fully structured way. So all of that html/markdown that we extracted from our deep crawl is finally useful now! You can imagine already how personalized email outreach can get just from our initial crawl and ingestion. But we can do way better! 11) Now let's see where we can get more signal: - We have job descriptions (which contain a ton of signal) - We have website information Now let's get social. This is honestly the fun part. But it'll have to wait til tomorrow. 😅

@BobbyThakkar probably @paysponge

software engineers and designers from India were already pretty good, but after the recent boom in agentic and AI coding, I am seeing some elite level talent come out of places like bengaluru, hyderabad, gurgaon and other cities. if they fix their venture capital ecosystem (very regulatory heavy & conservative on risk taking), they could see a huge potential boom - all of the talent and money is already there just waiting for a growth push.