signull labs exists to build @skye. but it also exists to acquire small companies & transform app first products into agentic experiences. excited to share that we closed our first acquisition today. more details soon, but these products will become core experiences inside skye. we’re also working on a major product update based on early feedback. very soon now. apologies for the delay on the beta rollout. the demand, messages, & inbound have been overwhelming in the best possible way. appreciate all the support. we’ll be much smoother & faster from here.

I really really think @signulll is onto something with Skye

organic feedback like this so motivating & deeply appreciated. we are ~30 days into the public journey & ~90 days into the overall product journey. & we have a lot of work to do to actually make a wonderful product that just works & also is cost effective with a good business model behind it. unfortunately we also are dealing with few hiccups like any small startup does. :( despite the vibe coding phenomena & the narrative around software being commoditized, building secure simple products backed up with a narrative that hits is still a very very very difficult task in a non deterministic computing era. the reason why building ai software is so hard is again the feedback loops aren’t as clean, you have to really *feel* it instead of simply model checking like in the previous era (i have made a ton of posts about this). we are working a ton to open the beta up to lots more peeps, appreciate everyone helping us get there.

okay i know i am biased af but having ai write to you on your home screen is a ~game changer. there is no better mechanic for ai to communicate with you & actually write ui (not walls of text). every future device will have this behavior. & we haven’t even done anything non obvious yet. i suspect it’ll be copied very soon.

whenever @skye analyzes my sleep for the week & gives me a report, it adds fun little trends in it too that add a level of analysis that is hard to do by the human mind. what is this if not a bicycle for the mind? this is just the beginning because your health agent should also suggest actions, & will have a much broader understanding of your context. all of it is completely ambient, promptless, & sorta fun as hell. i.e. my mom uses it.

1k beta users of @skye! we are letting in ~100 every day. great feedback so far. we are early & we have a fuck ton of work to do but god damn it is so much fun. i have never had more fun than building telling ppl why we even bother to do what we do. we don’t focus on outcomes, just refine our process. we are going to push a lot of boundaries on what it means to be ai native software that continuously runs & just works. we are only bottlenecked by time.

our goal with @skye is to build deeply *emotional* products that also provide real utility. this distinction matters a fuck ton because so many focus on pure utility. skye is definitively not a utility app (we leave pure utility to other ai companies). sure it gives you utility, but the product is designed to create emotional resonance through your life with music, photos, memory, place, surprise, & all the small fun things that make software *feel* alive. i can’t wait to share the agents we’re working on. as a tiny preview i asked some of my favorite spotify artist friends to compose original music for skye, like real music, made by real humans. now why the fuck would an ai company care this much about human made music?? you’ll see in the next release. music & emotional moments are going to be a huge part of what skye becomes. in the meantime, you can listen to our first skye album (vol 0) on spotify in the link below. my favorite tracks are somnia verde, infinity blue, & sun compass. we are only bottlenecked by time & ppl. please please join us, we are looking for amazing engineers… dm me with your work.

skye · welcome by @signulll

this is my current @skye home screen as i’m heading to bed. the feed ranking is getting incredible & we are expanding our agents capabilities rapidly. the future is agents simply telling you what you need to know when you need to know it with precision & action directly on a screen you glance at. going back to a screen full of apps feels like traveling back in time.

starting now…

consumer products need utility for sure. but utility alone rarely creates pull. the products that actually earn repeated attention tend to fuse usefulness with something more alive with things like play, delight, curiosity, aesthetic pleasure, & entertainment. most ai products today lean far too hard on pure utility & miss that second layer entirely. the winning ai consumer product likely won’t win on pure capability alone. it’ll win by merging capability with entertainment so tightly that using it feels less like labor & more like desire.

.@signulll says his new company is building "Facebook News Feed 2.0": a highly personalized feed powered by 22 AI agents that lives on your home screen: "We basically ask you to install two widgets — a medium widget and a large widget that encapsulates the entire home screen. And those work together." "The medium widget...shows you precisely what you might need to know at this point in time. And the big widget, what we call a For You widget, which is just a feed." "We're building the new iteration of the Facebook News Feed that's entirely AI-generated about your life. Highly personal, and that lives directly on your home screen. You can browse it as easily [as the feed]...We have 22 agents that work to generate content continuously for that feed."

honored to talk to john & jordi. i remember them reading some of my tweets very early & now they actually let me be on it??! god damn how do i lose? this is just the beginning for us. we are already putting the product into the hands of everyday users & we are iterating like hell. the future is background intelligence plus agentic action. the minutiae of your life, mediated & handled. that’s skye. our product anticipates rather than just responding. lots of stuff happening on our discord & we’ll continue to invest/grow out early community. if you’re an ios or android developer who wants to join our team, please dm me. you won’t regret the journey, i promise.

the bet for @skye is simple: your life should be processed & inferred through a personal intelligence layer without you having to do anything. every facet of your life from what you see, what gets done, what’s surfaced, & what’s silenced will run through ai by default with zero friction. & the ai will be responsible for presenting it through a variety of modalities (beautiful ux, push notifications, voice, video, visuals, & other ambient surfaces that we have yet to invent). it’s a monumental problem, & extremely difficult to get right but it will happen & you won’t know why you lived without it.

.@signulll says we're still in the stone age of AI tools: "I don't think most people are utilizing them anything beyond the basics." "We're in the stone ages of how people view and perceive and use these things, even though there's a billion people utilizing them, but they're not utilizing them to the full capabilities." "The number one challenge... is how do we make this stuff, the power of the models more easily accessible and useful in terms of what they can do." "I think this is happening with agents. It's happening today, but it still seems very primitive and very inaccessible to a lot of individuals... we need to make this stuff much more easily accessible and useful for individuals."

every time anish, erik & i meet up we end up in these weird often strange convos that stretch for hours. we often lose track of what we were even contemplating… sometimes it’s laughter, sometimes it’s something serious, sometimes it’s tech & culture & the underlying essence of kinda just being. i personally think these types of convos is what life is about. weirdly enough at the end someone always says oh my fucking god dude, what if we’d recorded that? erik is the master of this. so this time we did. & the other element was whether i’d show up anonymous or not. & since i’ve been leaning into the presence or whatever, it felt more way more fucking fun to lean into it. & the a16z crew whipped up this little fun avatar. it might be cringey or whatever. but the experiment at least feels interesting, esp for how we think about internet life vs. personal life. everyone has both now. anyway i might have said something regarded as always. but getting to do this alongside erik, anish & the a16z crew.. ppl who are without exception grace & class is an honor of the highest order. enjoy.

if you’d like to be an active member in our community & give us feedback, please join our discord. we will be doing ama’s, live previews, early access, give away’s, & a lot more! discord.gg/rWz9CGJQ

there are two unsolved problems in agentic commerce which is how agents pay for services they encounter & how users pay for agents that spend autonomously on their behalf performing tasks (buying things, paying for access to paywalls, etc.) our product touches on both of these & will expand these two primitives pretty fucking quickly. we will have premium agents that users subscribe to & all of our agents will eventually be equipped with a prepaid wallet they need to be able to function on their responsibility sets. at some point they will be able to spend money fully autonomously & report back. the internet is about to fundamentally change in so many different ways, it's absolutely mind boggling to think about.

we are starting to send out beta invites this afternoon! thanks for your patience! appreciate the overwhelming interest from everyone. we read every reply & dm! ❤️

i’ve been building ranking algorithms since the web 2.0 era. ranking is the defacto invisible infrastructure of modern life. it determines what you see, when, & why. but ranking in the age of llms is a fundamentally different problem, & almost nobody is talking about it. you see the old approach was statistical.. you needed labeled data, feature engineering, training pipelines, & constant tuning. if you wanted to deprioritize sentiment, that was months of work. now you can describe the ranking function in natural language & a model executes it. the bottleneck has shifted from engineering to editorial judgment. this is a categorical shift. ranking has moved from pattern matching on behavior to reasoning about meaning. the hard problem now is what should we mean by important. very few people are building with any of this understanding yet. for us, ranking via ai is one of the core infra’s of skye & we tune the fuck out of the ranking model through taste. i’ll post specifically about our feed & how it works because in essence we are kind of building a new type of ai feed for your life.