@bradmillscan To be fair they are not mature enough to be used to run a company... It's more of a fun sidekick than anything,...
English
Satyaa Goyal
1.6K posts
Satyaa Goyal
@satyaa_goyal
Casual surfer...
Rajasthan, India Katılım Ekim 2016
47 Takip Edilen25 Takipçiler
OpenClaw and Hermes are like digital Galleons and Frigates.
Powerful but they require a crew to maintain them.
Trying to run a company using these tools by yourself is a doomed voyage.
I need a crew if I’m going to keep using these tools.
English
This benchmark mirrors my day-to-day experience very well. The gap between GPT-5.5 and the rest of the field is massive, tangible, and very real.
Curious to see where Grok Build and Composer-2.5 would land here. What do you think?
Serena Ge (Datacurve)@serenaa_ge
Today we’re releasing DeepSWE, a new standard for agentic coding benchmarks. On public leaderboards, top models often look relatively close in capability. DeepSWE shows where they actually diverge, reflecting the realistic experience of developers in their day-to-day work.
English
@TheViableEdge Idk man... I think my experiment is better I had a spare phone number which I linked with mine Hermes agent and a seprate curated profile and added my family members access now they are talking with that ai agent and ngl the profile and the skills its generating are real gem...
English
I gave my daughter a Hermes agent + Gbrain, and told her to just use it, and to think of it as her second brain and personal assistant.
It's been a few days, but I'm purposely holding back on pushing her to use it because i want to see what happens organically.
My favorite part of this journey - I set up an email bridge from my agent to hers, our agents can email each other. I have hers deliver a health check once a week, and scan for emails from my agent a few times a day.
This allows me to send her agent ideas and can even set up cron jobs for her. I had her agent pro-actively use Excalidraw, to both surprise and make her aware of it.
English
Got that latest Qwen for ya in Hermes Agent now!
Nous Research@NousResearch
Qwen 3.7 Max is now supported in Hermes Agent
English
@blueprint_os @Voxyz_ai Ikr ntm these soul.md and other md files its better to just optimize those to clean 3-4 liner max only your main agent which is also your orchestrators should have in depth soul and he should be there for management role only...
English
@Voxyz_ai We run 5 agents off shared memory files. The startup context bill is real. Spent more time on what NOT to load than what to load.
English
>Karpathy just put Openclaw on stage at Sequoia as the software 2.0 example,
>his demo: you copy paste a prompt, the agent installs Openclaw on its own,
>but this paradigm has a bill, and most of it sits on what you loaded in,
>startup files, tool schemas, history bleed, memory bloat - i broke it into 7 receipts,
>glance at them every morning, a few weeks in you spot all 20 leaks, the bill keeps getting thinner.
Vox@Voxyz_ai
English
@Voxyz_ai To be fair the main reason for token wastage is relying on your agent way too much... And trying to do shit from one agent alone having multiple specific agent is the way to go... And set cheaper models or even free one for simple tasks...
English
@Sentdex I mean more then 80% of it jobs can be just a chill ao agent job 😜
English
Computer use is getting insane!
Geoff got himself a nice comfy white collar office job with a dell pro max workstation wtf
English
This is the new standard for engineering evals
Serena Ge (Datacurve)@serenaa_ge
Today we’re releasing DeepSWE, a new standard for agentic coding benchmarks. On public leaderboards, top models often look relatively close in capability. DeepSWE shows where they actually diverge, reflecting the realistic experience of developers in their day-to-day work.
English
@CuteCatsMagic How can he play games when he got such a cutie.
English
@KayvonJafar I think it should be a built in option so more peoples and non tech savvy can also get their hands on such cool tools..
English
hermes desktop is a big shift.
hermes agent just went from a terminal toy to something normal people can actually use.
this is how agents go mainstream.
English
@bradmillscan By default its a new born 😞 but do 2&3 session and then when auto memory sync skill build kicks in it starts becoming truly yours... In my case I had rly good openclaw setup so I didn't face those issues...
English
Hermes is just as lacking as OpenClaw when it comes to an operating system that you can do real work with out-of-the-box.
You're going to be working for days trying to setup a default Hermes to get it to a point where it's usable & helping you get shit done.
English
@Voxyz_ai That's way too fking long article now i am gonna have to waste more tokens to summarise it lol 🤣
English
@dedene True but to be fair copilot has become kinda corporate standard you know... And there generous limits had make people really lazy about there ai usage I have seen my co workers asking a mode server to be run by an ai.. complete handicapped....
English
@satyaa_goyal They got us completely hooked.
But competition from open source and models like Cursor's Composer-2.5 will force everyone to keep pushing limits and be as efficient as possible.
English
5 days left until GitHub Copilot switches to pay-as-you-go.
The free ride is officially over.
If you were using it, what is your move? Sticking around and paying up, or jumping ship to Cursor or Zed?
Theo - t3.gg@theo
I sent a single message on Copilot and it did over 60m tokens. It's still going. $30 of inference so far. In their current billing model, you get 1,500 messages, regardless of how expensive each is. I'm pretty sure I can do $45,000 of messaging on this plan
English
