Kevin Simback 🍷 @KSimback

COO @delphi_labs - building + investing in AI and crypto. Ex @IBM, @McKinsey, @CarnegieMellon, reformed CFA hermesatlas.com Puerto Rico Joined March 2015

Tweets

7K
Followers

19K
Following

904
Likes

27K

Kevin Simback 🍷 @KSimback

33 minutes ago

Great post on inference from my friend @0xSammy Give it a read!

0xSammy @0xSammy

7 hours ago

x.com/i/article/2069…

13 7 76 13K 36

2 0 6 571 0

View Details

If you’re not US or China, I don’t know how you sit back and become dependents on them for the intelligence that will define the future You’d think it would be a national priority to have a strategy here Of course, easier said than done - you need big $ and talent to compete with the frontier labs The easiest first step is to build sovereign data centers and host the open weights models or post-trained versions - at least you own some form of the intelligence The UAE and Saudi Arabia seems to be going this path and making moves towards training The most impactful next step would be focusing on decentralized training If you can’t build a cluster capable of training a capable model, decentralized training may be a viable option Not for today, but as an option to hedge the risk of being locked out or priced out of the US and Chinese models if that day ever comes Doing a deep dive on decentralized training with a report coming out soon Let me know if there are any projects you want to see covered, I’ve got a pretty good list already but open to include others if they’re legit

5 0 15 2K 1

View Details

Kevin Simback 🍷 @KSimback

6 hours ago

@bnivanovX Anyone operating at the forefront of AI is likely challenging their cognitive capabilities quite a bit, there’s a constant learning motion you have to sustain

2 0 0 83 0

View Details

Kevin Simback 🍷 @KSimback

6 hours ago

If you’re doing diligence on a new startup, you now need to assess their “AI nativeness” alongside team, product and market Why? If they’re not at the forefront of using AI, it puts them at risk of not executing fast enough It’s not a hard pass filter, but it is signal - here’s a few early tells: 1. Did they give you a .md file, alongside a deck, that you can feed to your LLM? 2. Did they build an interactive HTML deck or just send a PDF/docsend? 3. Do they talk about proprietary skills, data, evals or methods they’re using to build their product? 4. Do they ship significant product updates between meetings? Not an exhaustive list, just some anecdotal signs that I’ve noticed

5 0 26 3K 10

View Details

Kevin Simback 🍷 @KSimback

8 hours ago

@Takuiten It’s quite easy to filter out the larps from the serious ones

3 0 2 154 0

View Details

Kevin Simback 🍷 @KSimback

8 hours ago

@anildelphi @Shaughnessy119 🤝

0 0 0 291 0

View Details

Kevin Simback 🍷 @KSimback

9 hours ago

@milesdeutscher Agree on the cost-per-intelligence point, and I’m using it too via the coding plan sub But curious for you as an individual user of AI, how does the cost factor play in since you can get subsidized Opus or GPT via $200/month plans? To me it’s more of an enterprise cost factor

2 0 2 457 2

View Details

Kevin Simback 🍷 @KSimback

9 hours ago

@ACsstyle Only way to ensure he escapes the permanent underclass

0 0 2 72 0

View Details

Kevin Simback 🍷 @KSimback

10 hours ago

My 11yo son built a genuinely useful travel packing app as a PWA Now he’s asking me to get him an Apple developer account and saying things like “monetization”

4 0 33 3K 0

View Details

Kevin Simback 🍷 @KSimback

11 hours ago

@Teknium Have a feeling you’ll be hearing about many more businesses using it soon!

0 0 1 315 0

View Details

Kevin Simback 🍷 @KSimback

11 hours ago

@jasonzhou1993 Hey Jason, this is good stuff! You may also like a complementary tool I built called Looper which helps you design good loops and gives you a visual flow right in Claude/Codex to see how it works before you run it

Kevin Simback 🍷 @KSimback

5 days ago

Feeling a bit loopy? Introducing Looper - your loop design coach It’s been made clear that we shouldn’t be promoting agents, we should be designing loops You can use /loop or /goal but if your loop is poorly designed then it won’t produce good output Even worse, a poorly

24 34 419 195K 906

0 0 1 334 0

View Details

Kevin Simback 🍷 @KSimback

12 hours ago

@karsenthil Good feedback, thanks for sharing, and very familiar with Pluralis, I’ll be covering them in my decentralized training piece

1 0 3 103 0

View Details

Kevin Simback 🍷 @KSimback

21 hours ago

@lex_node 👀

1 0 1 895 0

View Details

Kevin Simback 🍷 @KSimback

22 hours ago

@Takuiten @IdleProtocol Thanks, will look into it

2 0 1 176 0

View Details

Kevin Simback 🍷 @KSimback

a day ago

The core premise for decentralized inference - it's a hedge against censorship of intelligence, whether that comes from governments or the frontier labs withholding it The additional value props like promises of cheaper tokens, verifiable inference and privacy are all secondary to this one core bet of censorship risk mitigation This report goes deep into the hard problems that need to be solved for decentralized inference to actually work - the technical challenges are hard and the market is unproven But there's a huge opportunity for any projects that can get this right, and it's not just serving inference - it's also about the data that can be used for the training-inference loop I profile a bunch of projects that I'm watching in this space including: @dphnAI @inference_net @MorpheusAIs @chutes_ai @c0mputeAI @tryParallax Darkbloom @eigenlabs @HyperspaceAI MeshLLM @TheBlockCo @usepod_ai @AntSeedAI @AskSurplus