←
Alexander Lin4,097 posts
Alexander Lin avatar

Alexander Lin

@linfluence

cofounder @reforge_vc

◷ Joined June 2009▧ 304 Photos and videos
Alexander Lin avatar
Alexander Lin@linfluence·3h

Is a new deployment in the works bc Together removed the Pearl inference endpoint back in August

Post attachmentPost attachment

Ryan Watkins @RyanWatkins_ 7h Article Pearl Thesis – Beyond the Bitcoin Standard For the first time in years, Bitcoin is being tested. Not in terms of price or legitimacy. But in terms of its long-term viability as the world’s preeminent digital store of value. In Bitcoin’s path

Alexander Lin avatar
Alexander Lin@linfluence·Nov 9

why does your graph mog mine

I use about 5-6 different inference providers across 3-4 harness. This does not include the web interface I use from time to time. What I find the most difficult to manage is my memory. They used to be fragmented to a point that I spent a lot of time finding the right conversation, copying pasting and all that. I even tried to roll my own "2nd brain". Obsidian vaults and all those "skills". None worked nicely because if its something I have to even spend a minute thinking/managing, then it simply isnt good enough for me. Also, as the chinese models improve i find that the most precious things are the years of conversations I have had with AIs and not the now commoditized "intelligence" that these models provide. The retrieval is HIGHLY important - it makes no sense for AI to waste context reading .md files and trying to figure out where things stand. A graph database is preferred here: About 6 months ago, Ava, our agentic sourcer notified me that someone is building this exact memory harness I was looking for: @nowledgemem (yes we are investors) Here is my set up: I've run every AI tool I use through @nowledgemem for months. 3,200+ conversations and 1,000+ memories later, here's what actually makes it great 🧵 One memory for every AI MCP server + CLI + plugins. I use it in Claude Code (@claudeai), Codex (@OpenAIDevs), ChatGPT (@ChatGPTapp), Grok (@grok), DeepSeek (@deepseek_ai), Grok Bot and Muse. Decide something in one tool, the next one already knows. No more re-explaining my setup. Its truly "hands off" - It keeps the actual conversation Full session threads are saved and searchable, not just AI-written summaries. "What exactly did we decide last month?" gets the real transcript, not a guess. - Search that finds both kinds of things Hybrid search: BM25 keyword + vector embeddings in one query. Finds the exact error code AND "that thing we talked about in October." - It knows a rule from trivia Every memory is typed (decision, procedure, preference, learning) and given an importance score. "Stop asking me routine questions" follows me into every agent as a rule, not a buried note. - Every chat starts already briefed A session-start hook loads a daily Working Memory briefing into the agent. New chat, zero setup. It already knows what I'm focused on. - Receipts for every idea Each memory links back to the tool and thread it came from. It once caught that a research idea came from an old ChatGPT chat, not from me. That's trust. - Local-first The server runs on localhost and the data stays on my Mac. You can bring your own model keys. My memory lives on my laptop, not in someone else's cloud. - reachable when I choose Remote MCP access protected by OAuth (mine goes through a Cloudflare tunnel). Even cloud ChatGPT can use my local memory, behind a login, without exposing my machine. - It draws its own map It pulls out entities automatically (6,400+ for me) and groups them into topics (45). My projects get organized without me tagging anything. Honestly I mostly use search, but it's a nice bonus. The big one: a real 2nd brain, because it's a graph Memories, conversations and entities are all linked. Mine has 6,400+ entities in 45 topics, built automatically. Agents follow the links to the right decision instead of reading .md files. Like my brain, everything connects. Last but not least: if an AI provider ever closes your account (important for my chinese fellows), you don't lose your history. Your conversations are already saved in your own memory.

Post attachment
Alexander Lin avatar
Alexander Lin@linfluence·Nov 5

the ocean is substantially underdeveloped relative to its economically achievable potential so i'm spending a ton of time looking at maritime/subsea tech co's plz reach out if you're building, researching, or just passionate about the subsea market

Alexander Lin avatar
Alexander Lin@linfluence·Oct 15

agentic credit made possible through @credifi

mod323 @itsmod323 Oct 15 Article Credifi Zero Introducing Zero Credifi Zero underwrites a crypto wallet with zero personal info. Every lender starts with the same question: who are you? Most of the time, the answer is a name, an address, a bank

Alexander Lin avatar
Alexander Lin@linfluence·Oct 12

notte team on a generational run if agents could get fomo, they would if they're not running browsers on @nottecore

Post attachment

Browser now costs $0.01 per hour on @nottecore 🖤 → Up to 12x cheaper and 7x faster! → one browser running all day: 24 cents → 1,000 browsers for an hour: $10 Ask your agents to run browsers on Notte: notte.cc/skill.md

0:20
Alexander Lin avatar
Alexander Lin@linfluence·Oct 12

chinese open weight models continue to prove they are good enough to be the building blocks for american ai products

Post attachment
Alexander Lin avatar
Alexander Lin@linfluence·Oct 9

have a team looking to rent on-demand 4-9 nodes of b300s (they don’t need to be connected) expecting 3-24 hours of usage and willing to pay market premium

Credifi avatar
⟳ Alexander Lin retweeted
Credifi@credifi·Oct 8

1/ Today we're launching our zkTLS verification iOS app and Chrome extension, starting with credit score screening from Experian, Credit Karma and NerdWallet.

mod323 @itsmod323 Oct 8 Article Cheaper loans with zkTLS The cost of information Unsecured lending runs on two things: information and enforcement. The better and cheaper a lender can assess a borrower, the cheaper and more accessible credit becomes. A

Alexander Lin avatar
Alexander Lin@linfluence·Oct 1

imagine being able to trade this with lev if you’re a macro pm or trader who wishes they could’ve (along with hundreds of other macro markets), lmk bc I know a team who unlocked this @forcexyzhq

BREAKING: The UK’s 30Y Government Bond yield rises to 6.02%, its highest level since January 1997. That’s a total of +564 basis points since the low seen in 2020.

Post attachment
Alexander Lin avatar
Alexander Lin@linfluence·Sep 30

conducted a survey across my non sv/pa network (~80 above avg persons thus far) to discover: 1. most have not heard of Instinct and if they have they don’t care bc they’re “not lazy” 2. if they did care enough to try it, immediate churn bc it “sucks and keeps messing up”

palo alto people are building agents for a world where everyone has 47 tasks at once while the average person is so fucking bored they’re doomscrolling during work hours

Alexander Lin avatar
Alexander Lin@linfluence·Sep 30

the early-stage vc barbell has shifted to investing in founders experiencing their own power law phenomena (only 1-2 out of the 30 things they've built are hits) and missionary founders working on problems so big and so hard that focused commitment on one thing is still a req

TBPN@tbpnSep 29

.@sama says the advice to pick one thing to build and pour all your focus into it no longer applies. "The canonical advice that was correct two years ago was, 'Pick one, and put all your effort into it, because you can't build all 30 things.'" "You can now build all 30 things. And you can get people to use them all. And you can work more on the ones that work." "So I think this mindset of lots of ideas, quick feedback, and being open to a lot of things not working is great for this new era for how people build."

0:57
Alexander Lin avatar
Alexander Lin@linfluence·Sep 24

who is ACTUALLY trading compute (ie, ocpi-h100-perp on architect or gpu rental prices on kalshi/pm) today?