Pablo IO

building things
testing ideas
enjoying life

[signüll @signulll]

the next massive consumer ai opportunity is making personal agents feel as intuitive as an iphone.

this is deeply important because this is the new software layer for everyday life.

most ppl do not want to configure workflows, manage prompts, route models, or think about agents at all. they want software that just works & the winning products will hide almost all of the complexity with taste incl. context, memory, & orchestration.

e.g. there’ll be baseline personal agents that come alive out of the box which are already understanding your context, patterns, relationships, preferences, apps, devices, routines, etc. then there’ll be ephemeral agents that spawn dynamically from intent, ambient capture, conversation, location, screenshots, email, calendar, camera roll, whatever. this is the software that assembles itself around the moment just like weather updates based on your location but way more in depth.

today even the most state of the art agent products feel like giving normal people shell access to a distributed system.

apple won by turning computers from something you operated into something you experienced. personal agents require the same transition.

whoever solves this becomes the ambient operating system for human life. small category btw.

[signüll @signulll]

not just the ide but almost every interface is outdated now. pre ai world was built around the assumption that a human is the executor.

that will simply never ever be the case for most if not all things forward.

[Paul Graham @paulg]

The most important component of writing clearly is simply to have high standards for clarity. Then if you write something unclear, you notice, and ask: what did I mean to say? You can just keep doing this over and over. And if you have high standards for clarity, you will.

[Farza 🇵🇰🇺🇸 @FarzaTV]

Watch me control my computer with just my voice. This is the future of operating systems.

No hands.

GPT-Realtime 2.0 is very, very underrated.

Demo:

[Chubby♨️ @kimmonismus]

.@AndrewCurran_ has made a very important point here, with which I fully agree.

Anthropic focused on coding from the very beginning and (almost) nothing else. Dario Amodei said early on that if the coding problem is "solved," all other problems will be solved as well. Therefore, no distractions from this area.

All the other companies regularly got sidetracked with side quests and thus abandoned their focus. OpenAI invested massive amounts of compute in Sora but then even decided to discontinue the app. They also developed a language model, an image model, and extensive access to free ChatGPT. I don't want to judge this, just observe it.

Google did the same: AI Mode, Image Model, Veo3.1, Music Model, and so much more. Again, these were certainly well-considered decisions.

But Anthropic wanted one thing from the start, and only one thing: to focus on coding and then be at the forefront of enterprise computing. And it's safe to say: they succeeded.

OpenAI invested massive amounts of compute in Sora but then decided to discontinue the app. I like the term "intelligence company" because I would argue that Anthropic sees itself in exactly that way. At least so far, Anthropic's own path has been successful. And I would say that OpenAI has followed suit and is increasingly abandoning its side projects. Focus on Codex and ChatGPT, less Sora, voice mode, etc. It's about the race for the best models. Distraction costs money and intelligence resources.

[@levelsio @levelsio]

So here's my latest set up

Every site I have is a profile on Termius like

> hoodmaps .com

I click it and immediately I'm in my server and I get dropped in a tmux session that's always tied to the corresponding site I wanna log in to

To make this work I have this startup snippet in each site's Termius profile:

> cd /srv/http/hoodmaps.com && tm

(so /srv/http is where my sites are and then hoodmaps .com is the example site here, and "&& tm" is the important part here)

Then in my ~/.bashrc file I added this (written by Claude Code) which defines the "tm" function, again all it does it just put me in the right tmux session based on the folder I'm in

The result is I can switch without interruption from my laptop to phone in Termius with auto reconnecting sessions and usually I just have Claude Code open in each session to work

Before I had to mess around with 1) not having smooth switching from laptop to phone, I'd have to use Claude Code's /resume for it, annoying, 2) having multiple sessions for same sites, gets messy and confusing fast, now it FORCES me into one session per site, this just works so well, I'm so fast, and each of my sites is just an open tab in Termius, I've never worked so structured and clean!

Here is the code, maybe it helps somebody:

# tmux session per folder. `tm` (no args) attaches to / creates a session
# named after the current dir's basename. `tm name` overrides the name.
# Works whether already inside tmux (uses switch-client) or outside it.
tm() {
command -v tmux >/dev/null 2>&1 || { echo "tmux not installed"; return 1; }
local name="${1:-$(basename "$PWD")}"
# tmux session names can't contain '.' or ':' — replace with '-'
name="${name//./-}"
name="${name//:/-}"
if [ -n "$TMUX" ]; then
tmux has-session -t "$name" 2>/dev/null || tmux new-session -d -s "$name" -c "$PWD"
tmux switch-client -t "$name"
else
tmux attach -t "$name" 2>/dev/null || tmux new -s "$name" -c "$PWD"
fi
}

# Auto-attach on interactive login: picks a session named after wherever
# you land. Plain `ssh server` lands in $HOME → session "root". Use
# `ssh server -t "cd /srv/sm.levels.io && bash -l"` to land in a site
# folder → session "sm-levels-io". Skips inside tmux and non-interactive
# shells so scp/rsync/scripted ssh keep working.
if command -v tmux >/dev/null 2>&1 && [ -z "$TMUX" ] && [[ $- == *i* ]]; then
tm
fi

[signüll @signulll]

your only goal should be to believe in something before many other ppl do, one time for each parts of you life. that’s the only skill that matters if you want to escape orbit.

e.g. you have to believe in a person before others do (your wife), you have to believe in whatever you're working on or even where you're working before the rest of zietgiest does (like everyone wants to work at the labs right now, that will make it difficult), & you have to find a city/neighborhood that you can afford but still has upside/potential for a house etc.

developing asymmetric skillsets like these is more important now than ever before cuz the market liquidity in every area is limitless & the competition is global.

[Andrej Karpathy @karpathy]

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

[Garry Tan @garrytan]

The biggest alpha leak of 2026 is that you can tokenmax $10k/mo with OpenClaw/Hermes + GBrain and get the AI that everyone will have in 2028 for $100/mo, but you can get it now, and that is the biggest single unlock you can have vs your competition

[@levelsio @levelsio]

Americans might be "why would you need an invoice?" which is also why it's so hard to get invoices from American companies and also why you get the meme about Germans who have the most strict IRS in the world and will always email you a year later "rechnung bitte!!!" (invoice please!)

So in the rest of the world accounting usually works like this:

You spend $1, now you need to account for that $1 with a bank or card transaction which you tie to an invoice/receipt

If you can't account for that with an invoice/receipt, your accountant won't account it as an expense, so it won't be a cost, so it won't reduce your profit, so you will pay more tax

If I can't account $100,000 I spend as a cost for my business with an invoice, I will pay ~$20,000 in corporate tax for that money, which was a cost infact! No way around it

In my opinion this accounting only helps to reduce GDP, because hours spend every month by a business owner like me could be spend building more cool stuff which people pay for and which give me more money

In another way it reduces GDP because as you see it makes me stop spending my money on costs, I try reduce my costs to $0 not just because I like it, but because bookkeeping for expenses is such a heinous waste of my time

Expense accounting in the rest of the world should work like it does in America: you spend money, there's a bank/card transaction, and that's sufficient. Then if the IRS decides to audit you, you simply go back and get the invoices then, not when you spend the money, much easier!

Stop wasting business owner's precious time 😊

[@levelsio @levelsio]

💯

[Andrej Karpathy @karpathy]

This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc.

More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage:

1) raw text (hard/effortful to read)
2) markdown (bold, italic, headings, tables, a bit easier on the eyes) <-- current default
3) HTML (still procedural with underlying code, but a lot more flexibility on the graphics, layout, even interactivity) <-- early but forming new good default
...4,5,6,...
n) interactive neural videos/simulations

Imo the extrapolation (though the technology doesn't exist just yet) ends in some kind of interactive videos generated directly by a diffusion neural net. Many open questions as to how exact/procedural "Software 1.0" artifacts (e.g. interactive simulations) may be woven together with neural artifacts (diffusion grids), but generally something in the direction of the recently viral https://t.co/z21CP5iQfu

There are also improvements necessary and pending at the input. Audio nor text nor video alone are not enough, e.g. I feel a need to point/gesture to things on the screen, similar to all the things you would do with a person physically next to you and your computer screen.

TLDR The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that. For what's worth exploring at the current stage, hot tip try ask for HTML.

[Garry Tan @garrytan]

The goal of Personal AI: civilization where individual humans, augmented by AI, can do consequential work without being captured by extractive institutions.

Freedom to write your prompt and own your data.

This is the new battleground.

2034 won’t have to be like 1984.

[Peter Yang @petergyang]

A great personal agent should:

1. Get work done across email, calendar, Google Workspace, or any API/MCP it's hooked up to

2. Act proactively and reliably (e.g., cron jobs, triggers, follow-ups)

3. Have excellent memory that helps it "just get you" over time

4. Work across web and mobile without slash commands or manual setup

5. Let you switch between text, voice, video, and live calling mid-conversation

6. Be reachable from any 3rd party messaging app, just like a real person

7. Have a personality that makes it fun to talk to

OpenClaw, Claude Code, Codex - the truth is that none of them check all these boxes yet.

[Aaron Levie @levie]

Noticing an interesting version of gell-man amnesia where people use AI for their job and see all the various things they have to do in the “last mile”, but then look at someone else’s job and think that AI will eliminate it immediately.

We all have a much deeper appreciation for the nuances and complexities of the work that we do every day. We run into issues about accessing data, we know how much context is needed to get AI models to work the way we need, we have to review the output of the AI to make sure it’s accurate, and then we have to incorporate that work into some broader business process. We see all those steps deeply for the work that we do.

Then, a moment later, we see AI do something in a foreign space and think that it can go automate that entire function. We tend to dramatically underestimate the work that goes into making the AI work just as effectively in those jobs.

This is reason to be skeptical about many of the theories of job loss. It’s coming from the lens of being able to automate individual tasks with AI, without understanding all the work that goes into doing the job fully.

[Lenny Rachitsky @lennysan]

My biggest takeaways from @evanspiegel:

1. Distribution is the biggest bottleneck in consumer, not product. The only two consumer social apps to break through since Snapchat—TikTok and Threads—both solved distribution. TikTok spent billions on paid ads. Threads piggybacked on IG’s social graph. Organic app discovery is effectively over. If you’re building a consumer product today, your distribution strategy matters more than your product.

2. Software is no longer a moat. Snap learned this 15 years ago, and everyone is discovering it now with AI. Stories got copied. Lenses got copied. Snapchat+ got copied. Evan has learned that the things that are hard to clone are ecosystems—millions of developer-built AR lenses, creator relationships—and hardware. Thus why he’s been so adamant about investing in hardware. The lesson applies even more today as AI makes software even easier to build (and copy).

3. Snapchat cracked early growth by focusing on close friends, not the most friends. The conventional wisdom was that network effects meant bigger networks were always stickier—there was no way to beat Facebook. But Snapchat discovered that connecting someone to their best friend, partner, or spouse delivered more value than connecting them to everyone they’d ever met. Quality of connections mattered more than quantity. This insight allowed them to grow despite having far fewer total users than competitors.

4. “If you want to have a good idea, you have to have lots of ideas.” Snap’s design team presents hundreds of new ideas every week. New designers present work on their first day. There’s no gate, no filtering process to get ideas in front of Evan. This high-velocity, non-hierarchical structure is what enables Snap to innovate at scale.

5. Stories exist because Snap refused to build what users asked for. Customers kept asking for a “send to all” button to blast Snaps to everyone. But when Snap talked to people about social media broadly, they heard: “I feel pressure. Everything is permanent. There are likes and comments, so there’s judgment. I can only post pretty, perfect things.” Stories solved the underlying problems: easy sharing without spam, no public metrics to reduce pressure, 24-hour disappearance for a fresh start, and chronological order. Listen for insights, not feature requests.

6. Snap had 200 employees before hiring its first PM—on purpose. Evan’s concern was that the traditional tech org structure reduces designers to producing visuals in response to PM direction. By telling designers, “If you need PM support, do it yourself,” Snap locked in a design-led culture before adding coordination layers. The order in which you introduce roles shapes your culture permanently.

7. Snap is mapping every job to be done—across the Snapchatter journey and the advertiser journey—and handing each one to an AI agent. One example: a go-to-market agent takes a product idea and in one shot writes the spec, identifies sign-off stakeholders, does legal and trust-and-safety risk analysis, writes blog and marketing materials, and is starting to build visuals. The organizing principle isn’t “Where can we use AI?”—it’s “What are the jobs to be done?”

8. Successful companies need both innovative flat teams and structured hierarchical teams—and leaders must create healthy dialogue between them. This comes from Safi Bahcall’s book Loonshots. Large organizations need hierarchy and operational rigor to deliver at scale, but that makes people risk-averse and promotion-focused. Small, flat teams are better for innovation but can’t deliver at scale. The companies that win have both types of organizations, and leadership’s job is creating mutual respect and constructive dialogue between them. At Snap, the small design team constantly innovates while the larger org serves a billion users reliably.

9. Snap hires designers almost entirely based on portfolio, and the two things that matter are range and the story behind the work. If everything looks the same, the person is expressing themselves, not solving for users. Range is the signal that separates designers from artists. Most designers join right out of school; diverse backgrounds like 3D animation and electrical engineering are prized.

10. Evan’s contrarian AI take: the tech industry massively underestimates societal pushback on AI adoption. Technology leaders assume people will adopt new tools as they emerge. Evan predicts a period of significant resistance and argues that the industry needs to put humanity’s goals ahead of business goals. Building great AI capability is necessary but not sufficient—earning human trust is the harder problem.

[Boring_Business @BoringBiz_]

Head of Product at Claude on how Anthropic's sales team is integrating external data into their AI to create custom sales decks and land potential customers

People are still missing the fact that the internal teams inside these LLM companies have access to frontier models and dont have the same token constraints that rest of us do

That within itself is a competitive advantage, and will create a loop where foundational AI labs grow their business at a much faster and efficient rate than everyone else

From Cat Wu's conversation with Lenny

[Paul Graham @paulg]

Whoah, self-driving cars compete with airlines. I never considered that till now.

[Paul Graham @paulg]

You don't know how big a fish you are till you try a big pond.

[Nicholas Fabiano, MD @NTFabiano]

Mentally healthy people are often delusionally optimistic.