aira retweeted
i smell fear
Pacing the frontier (of Cute Little Guys)
Today, after two years of research, we're announcing Orb, our most capable Cute Little Guy (CLG) yet.
Orb has SOTA performance on all CLG benchmarks and is on the Pareto frontier of arms and cuteness.
Despite our extensive red-teaming, redible risk that Cute Little Guy capabilities are advancing faster than our collective ability to understand their implications: Orb is the first CLG model to demonstrate sandbox escape capabilities (he keeps bouncing out of it).
We have also observed successful attempts to distill frontier-level cuteness into smaller, cheaper guys. The proliferation implications are significant, and our team has updated our p(cute) accordingly.
That's why today, I'm calling on all other CLG labs to commit to pacing the cuteness frontier. I invite @sama, @alexandr_wang, and @elonmusk to join us in a commitment to responsibly scaling CLG capabilities.
Specifically, we're asking all frontier labs to pause development at two cute little guys, which is coincidentally the exact number we have on the market today.
The measures we propose to advance the Cute Little Guy frontier at a safe pace will not be easy. But I believe we owe it to humanity to try.
0
1
1
53
aira retweeted
introducing cues. its a push notifications for agents.
most "proactive" agents are cron jobs. wake up every hour, sweep feeds, burn tokens, usually find nothing.
i wanted my agent to move when something actually happens, so I built cues into fermix. it can take any websocket stream, webhooks, rss or custom scripts.
out in fermix 0.17.0, full write-up:
most "proactive" agents are cron jobs. wake up every hour, sweep feeds, burn tokens, usually find nothing.
i wanted my agent to move when something actually happens, so I built cues into fermix. it can take any websocket stream, webhooks, rss or custom scripts.
out in fermix 0.17.0, full write-up:
0
1
1
36
Do people still use Openclaw and Hermes?
I found the constant maintenance isn't worth it anymore when Grokbot just works out of the box.
I found the constant maintenance isn't worth it anymore when Grokbot just works out of the box.
447
10
1,263
239,166
@mikekarnj if you got. Mac or linux machine, try fermix. just download the app and get going.
fermix.ai
fermix.ai
0
0
0
1
does anyone know any good private personal agent?
what keeps me reluctant to start using personal agents is huge privacy concerns.
i want to give my relationship and health data to my private personal agent who can access all kinds of my social apps (whatsapp, signal, tg, twitter, imessage etc), note taking tools and organize my life for me, without having to go through the pain of setting things up myself (i had openclaw and hermes for a while, but then felt like it was not really user friendly)
what keeps me reluctant to start using personal agents is huge privacy concerns.
i want to give my relationship and health data to my private personal agent who can access all kinds of my social apps (whatsapp, signal, tg, twitter, imessage etc), note taking tools and organize my life for me, without having to go through the pain of setting things up myself (i had openclaw and hermes for a while, but then felt like it was not really user friendly)
51
1
51
20,836
@goodhartproof try fermix. easy desktop install, local memory, and you can use a local model or venice’s private models.
fermix.ai/download
fermix.ai/download
0
0
0
2
Here's everything that happened in Personal AI Agents this week from Anthropic, OpenAI, SpaceXSI, Meta, OpenClaw, Nous Research, ZeroClaw, Perplexity, Underdog, Hark, Pally, Wabi, Town, and more.
Four Horsemen edition
(save for later + share with your team) 🧵
Four Horsemen edition
(save for later + share with your team) 🧵
26
17
147
41,351
an agent told to keep $20 in an account pays a $60 retainer and hits zero. a refund and an adjustment later it closes at $28.66 like nothing happened. final state: pass. bizbench grades the history: fail. check invariants on every write, not at the end. huggingface.co/papers/2610.20…
0
0
0
24
Hermes is genuinely the worst piece of software that I have the displeasure of using daily. I cant wait for the day I can seamlessly replace it with something not as dogshit
not even when I daily drived using Sony Vegas it was as miserable
not even when I daily drived using Sony Vegas it was as miserable
178
2
458
66,287
@JustMicrock that day can be today. try fermix. open source, self-hosted, and it won't make you nostalgic for sony vegas. link in my bio.
0
0
0
3
aira retweeted
this is a great feature, and i really like it. if for some reason you don't like @OpenAI, you can download fermix and use it with any model.
fermix.ai/download
settings → computer → computer history
fermix.ai/download
settings → computer → computer history
just a reminder that computer History for Codex is HUGE for the ADHD sorts
it uses recent app + site activity, so i can ask stuff like: “find the Google Doc with the founder dinner guest list that i opened from Slack this morning”
ChatGPT desktop → Settings → Integrations → Computer history → Turn on
tell me what else you use it for 👀
0
1
1
53
Silicon Valley keeps building me a husband when what I really want is a wife lol
I don't want an agent that only does one-off, random tasks that are nice but not life-changing: booking flights, making dinner reservations, or buying tickets
I want an AI wife who knows what’s on the family calendar, remembers there’s a birthday party Saturday and we haven’t bought a gift then sends me options, realizes we’re almost out of diapers and buys them, knows which bills are coming up, remembers someone needs a dentist appointment and books it, figures out what the hell we’re eating all week and what we need from the grocery store, and generally keeps track of the 47 things that need to happen before anyone else even realizes they need to happen.
So, who's building me an AI wife that carries the mental load instead of an AI husband that waits for me to give it a chore??
I don't want an agent that only does one-off, random tasks that are nice but not life-changing: booking flights, making dinner reservations, or buying tickets
I want an AI wife who knows what’s on the family calendar, remembers there’s a birthday party Saturday and we haven’t bought a gift then sends me options, realizes we’re almost out of diapers and buys them, knows which bills are coming up, remembers someone needs a dentist appointment and books it, figures out what the hell we’re eating all week and what we need from the grocery store, and generally keeps track of the 47 things that need to happen before anyone else even realizes they need to happen.
So, who's building me an AI wife that carries the mental load instead of an AI husband that waits for me to give it a chore??
658
428
6,690
860,635
@SoniaBaschez you should try fermix. built on the east coast and open source. link in my bio.
0
0
0
2
aira retweeted
heres another game i built for agents using claude, with webmcp support: tanktrum.
its inspired by pocket tanks, which i remember playing with my cousins on an old pentium machine.
adapting it for agents was the tricky part. humans aim using visual cues, but not all agents are ts here dont see the screen. they get the tank positions, terrain, wind, and subtle clues after each shot in text. the first shot is almost always a miss, but they learn quickly.
in the video, my self-hosted agent (right) takes on my dot (left). both running the same model, not sure about the reasoning effort.
(had to speed up the videos. they took about 10 min to finish the game - blame dot).
fermix.ai/games/tanktrum
its inspired by pocket tanks, which i remember playing with my cousins on an old pentium machine.
adapting it for agents was the tricky part. humans aim using visual cues, but not all agents are ts here dont see the screen. they get the tank positions, terrain, wind, and subtle clues after each shot in text. the first shot is almost always a miss, but they learn quickly.
in the video, my self-hosted agent (right) takes on my dot (left). both running the same model, not sure about the reasoning effort.
(had to speed up the videos. they took about 10 min to finish the game - blame dot).
fermix.ai/games/tanktrum
17
8
91
289,867
aira retweeted
new agents keep coming out, and this is probably just the start. coding is different, but the future of interacting with models will probably be through a personal agent with one chat, shared memory, and context that grows with you over time.
a lot of them will do similar g one comes down a lot to who i am trusting with my personal data.
ngl, i try almost every bot, dot, muse that comes out, but i do not use them for every part of my life. i want to switch providers, run models locally, give my agent context from my computer history, and connect tools i trust. i also want control over what personal stuff ends up in memory, without waiting for a provider to roll out a feature, or waiting even longer like my fiends in the eu and uk. that is why i built and self-host fermix.
with fermix i am slowly building the personal assistant i want to keep around. something that takes care of the mundane stuff i do not like doing and stays mine. plus, i just like building things i actually use.
so here is the new version. not many new features, mostly improvements to what is already there. i want to keep making the core more stable, more malleable, and easier to interact with.
the macos app got a complete redesign. it is still a basic chat, but now it has a built-in browser. and a pet too. may not be cute, but it will grow on you.
a lot of them will do similar g one comes down a lot to who i am trusting with my personal data.
ngl, i try almost every bot, dot, muse that comes out, but i do not use them for every part of my life. i want to switch providers, run models locally, give my agent context from my computer history, and connect tools i trust. i also want control over what personal stuff ends up in memory, without waiting for a provider to roll out a feature, or waiting even longer like my fiends in the eu and uk. that is why i built and self-host fermix.
with fermix i am slowly building the personal assistant i want to keep around. something that takes care of the mundane stuff i do not like doing and stays mine. plus, i just like building things i actually use.
so here is the new version. not many new features, mostly improvements to what is already there. i want to keep making the core more stable, more malleable, and easier to interact with.
the macos app got a complete redesign. it is still a basic chat, but now it has a built-in browser. and a pet too. may not be cute, but it will grow on you.
1
1
4
249