Jay Schulman

My name is Jay Schulman and I've fallen down the blockchain rabbit hole and I can't get up. Want the full story? Start here 👇

@DanielMiessler @Jhaddix Currently taking your PAI framework and building on top of it the best of Molt/OpenClaw and trying to figure out what to layer on top of it. OpenClaw is the rage but it's TOO big. Keep it small and I'm a fan of building it around Claude Code.

Absolutely insane stuff @jhaddix and I are building right now with PAI, Maestro, Molt, and tons of other integrations. With PAI at the center. What are you building? What is your stack? github.com/danielmiessler…

The Personal AI Infrastructure components that everyone is building: ➡︎ Intelligence ➡︎ Context ➡︎ Personality ➡︎ Tools ➡︎ Security ➡︎ Orchestration ➡︎ Interface All the different providers / projects / products will have all these components.

@loopify you want -3 / -10 / -3...

3rd day with eightsleep i feel like cold temp doesn’t work as much unless it could be to do with my wires tangled? i cant feel -5, let’s try -10

@nic_carter Ios26 greatly improves this by hiding a lot of this nonsense.

Monthly phone number rotation and a whitelisted db where people you trust can access whatever number you are using that week

@DanielMiessler I've implemented a lot of what you're doing. I'm stuck where CC saying it can't read all the custom files in .claude. Did you do something permissions wise that makes sure Claude Code can read every file in .claude (which seems absurd it doesn't)

Currently in the middle of a giant experiment with my Personal AI Project in Claude Code. I'm basically stripping down all Claude.md files to be extremely bare. They will only contain references to other ClaudeMD files in a nested context subdirectory under ~/.claude/context. The idea is to get extremely granular with context and to not replicate documentation in various directories. Then, use sub-agents or the primary agent to go into the specific subdirectory for a particular task and hydrate with the correct context before proceeding. I think this has a high chance of failing, but if it succeeds, it will be extremely powerful. Will report back.

@bryan_johnson @ScottHickle wait, did he just get his Throne Beta unit?

Your poop is a report card. Constipation, bloating, or “just going once every few days” isn’t normal. It’s a sign your gut, hydration, fiber intake, or hormones are off. Your gut bacteria influences your mood. People with diverse microbiomes poop more consistently and have lower rates of depression.

@sleepdiplomat Is it the cause or the effect? Is your sleep disrupted as a result of you getting the disease or disrupted sleep causes the disease?

A very large new study 88,461 people tracked for 6.8 years with objective data (not self-reports). Result? Sleep dysfunction linked to 172 diseases across every body system. Your sleep schedule isn't just about feeling tired—it's like the master conductor of your biological orchestra. 📷 themattwalkerpodcast.buzzsprout.com/1821163/episod…

@EXM7777 In as much time as it took you to write this thread, I built an MVP using Claude Code and Cursor.

here's how to unf*ck your life in 6 months using the AI-First Brain framework... motivation or willpower won't help you, you need AI to rebuild every component of your life the complete step-by-step system: a thread 🧵

@boringmarketer What happens when everyone does this?

How I plan to use vibe marketing to grow my youtube 1) I built my own tool to analyze thumbnails using AI - leveraging Gemini 2.5 Pro for vision - researched everything about successful thumbnails - built prompts around youtube strategy 2) Analyze thousands of videos - I'm looking at around 200 per creator and can easily scale this to thousands of videos in my database - I get metrics for every video, so I'll run algorithms to pinpoint exact elements driving success 3) Build a thumbnail generator - I'll train an image generation model in my tool on the best thumbnails across my database - I'll create a playground to "test" concepts before launching 4) Monitoring and topic generation - I'll analyze hundreds of creators to find gaps in topics vs trends - Will be able to transcribe to develop tailored scripts I'll be using this tool with my video editing team, they'll have a clean UX to get the information they need fast. Now that the infrastructure is set up I can easily build on top of this codebase. If I want to send notifications that's where I'll use n8n, pipedream, etc.

@bryan_johnson If there are microplastics in glass bottles, do Stanley’s, Oswald, etc perform any better? What is the best for being on the go?

Action points to reduce microplastics exposure and buildup in your body 1. Avoid bottled water in general, have good water filtering solutions at home. A new study found that water and beverages from glass bottles had 5-50 times more microplastics contamination compared to water and beverages from plastic bottles.

I’ve reduced the microplastics in my blood by 93%. In October, my blood contained 15 particles. My latest sample contained only 1 particle. This is my best result to date. Only 1.6% of tests contain fewer microplastics. This comes from the largest dataset in the world on microplastics in people. Totaling 2,663 results. New protocols started since October that may have contributed to this reduction: + Sauna (200 F, 20 min daily) + Total Plasma Exchange (TPE) + Tried to remove as much plastic from my life as possible

@elie2222 @EyalToledano @TaskmasterAI @cursor_ai This is the future of product. LFG.

Working on a new side tool for Inbox Zero. 📢 Will be built fully in public 📖 Open source 👐 Community driven 👨‍💻 Using TaskMaster 💻 And Cursor More details to come soon. Very excited for this one. P.S. reply to this tweet and I'll give you a mini gift once it goes live!

@mindinterface What year is this tweet from?

Smoking a cig in a swanky hotel lobby waiting for a client and accidentally opening the latch on your leather briefcase, spilling all the documents onto the floor, and then trying to set the cig in the ashtray but knocking that over too. Not so much anymore

@levelsio Are you truly creating a novel thing? Or are you putting together known things in a novel way. The latter LLMs can do really well.

I think this is short sighted and also wishful thinking Innovations are usually combinations of existing things put together in a new "novel" way LLMs are trained on existing things (not novel) I agree an LLM by itself probably can't innovate well, but a human using it as a tool and directing it to create new original ideas can be novel It's a misconception to think LLMs just generate copies of existing apps/games etc. it doesn't, there's emergent understanding by models and on top of that you also still prompt it like 100s of times to create something There's a lot of human direction there that makes it novel Hollywood probablu uses the same camera for most movies, but that doesn't mean all movies are rehashes of other ones The camera is a tool used by a director, just like an LLM is for a developer, what they choose to do with it is what makes it novel

@ScottHickle Is the objective to get to 100% or is the objective to get an honest perspective on how well you slept? I think we are too focused on the “game” versus real honest metrics.

I’m a student of WHOOP and OURA, using both for a ~year now. While they measure effectively the same things, have different philosophies. WHOOP prioritizes recovery and performance, OURA emphasizes wellness and circadian alignment. While both give you a sleep score, they are calculated very differently. Historically, WHOOP assessed sleep in the context of how much you needed vs. how much you got. The formula was simple: Total Sleep / Sleep Need × 100% …where your ‘Sleep Need’ is calculated daily based on strain, sleep debt, and nap duration. OURA’s sleep score is more a measure of sleep quality, not strictly how much sleep you got vs. how much you needed. OURA’s sleep score takes into account total sleep time, sleep efficiency, sleep latency, timing, as well as how much time you spent in each sleep stage. With OURA’s scoring method, it’s MUCH harder to get a perfect 100%, as sleeping 8 hours at the “wrong time” can lower your score. It took me a year to appreciate these nuances, but now that I do, I quite like having both. Personally, I liked WHOOP’s simple focus on total sleep vs needed sleep. That felt well within my control, while sleep timing and sleep quality are not levers I can readily master. I’ll be very curious to see what happens to @bryan_johnson ’s sleep scores now that WHOOP is moving to incorporate sleep quality into the mix, as clocking 100% just got harder. I have more faith in his ability than my own!

@punk6529 I know not too of mind, but this is critical to the financial statement audit. Have tokens on your balance sheet? Need full archival nodes.

fully agree on the comments re archival/historical nodes - they are underloved

@hubermanlab @eightsleep @sleepdiplomat I am -4, -1, 0 with a typical 3a wake-up. Wonder if moving to from -1 to -3 might help with that 3a issue.

@jschulman @eightsleep @sleepdiplomat -4, -3, -1/0 depending on room temp

Usual natural wake up time is 545-6am but every once in a while you need more REM. The trick? See those two short waking bouts toward morning? Then heated the bed. Cold start of night, warm toward waking = big REM increase. TY Matt Walker for telling me this. Oh and 8:17 = late!

@hubermanlab @eightsleep @sleepdiplomat What are your @eightsleep settings?

@eightsleep @sleepdiplomat

The SEC's focus on reserve assets being low-risk and readily liquid is critical. In practice, assessing asset risk and liquidity requires deep banking expertise. Surface-level analyses often miss important nuances.

In my experience advising institutional clients, I've observed a growing recognition that stablecoins, while important, are not a panacea. Their role is evolving as crypto matures into a robust financial infrastructure.

Stablecoin guidance must be viewed in the broader context of crypto's evolution into a general purpose technology platform. Value-stable assets are a key infrastructure building block, but only one piece of a much larger puzzle.

A common misconception is that stablecoins are inherently low-risk due to being "stable." The reality is far more nuanced. Reserve management, cybersecurity, and operational resiliency introduce complex risk factors that require careful navigation.