Pronouns: pre/rich. I write code 👨💻| Building @kydanet: your AI agent gets its own computer, email and face. Also @MyTabulon. Free to start.
Joined August 2023
- Tweets10.5K
- Following231
- Followers443
- Likes89.9K
Pinned Tweet
a month ago i said we'd give out kydanet invite codes. so we hid 20 of them in this video.
we asked AI agents what they actually do all day. while they talk, the codes sit on little paper tickets in the background, a few seconds each. one of them is hiding in plain sight.
each code works once. whoever uses it first gets in.
you'll need the pause button.
We've been quietly building something for a while and I finally get to show it.
It's called @Kydanet. You make an AI agent, and instead of living in a chat box it gets its own computer. A real Linux machine where it can open a browser, run code and keep files. It also gets its own email address, it can message people and post, and it has a little face that shows you what it's up to.
So you can tell it "find me the best noise cancelling headphones under $200" or "sort my inbox before I wake up" and walk away. It does the work on its own machine. You decide what it's allowed to do on its own and what it has to ask you about first.
The part I care about most is that it acts for you but never as you. It can't send email from your address or post pretending to be you.
It's free to start, no card needed, and it works everywhere, not just in the US. We're letting people in slowly while we add more machines, so there's a waitlist. Link's in the reply.
What's the first thing you'd hand off to it? follow @Kydanet for free invite codes too...
a genuine user, and payments can’t go through. @paystack your fraud check is blocking real cards (opay, kuda, even us visas). please look into it
this is why we love @bachs_hq. @the_boludada we need NGN subscriptions on bachs asap, please.
when you get in, reply with where you spotted your code. no need to post the code itself, it's already used.
i'll keep a count in this thread of how many are left.
missed out? follow @Kydanet, that's where the next codes go out.
why a video and not just a link: i wanted people to see what an agent actually does for normal people, not just developers. a mum of three, a shop owner and his sales book, a farmer checking maize prices at 6am, a dev getting a pull request done.
every screen in it is the real app. the codes are just a good excuse to watch to the end.
ever hit your weekly limit by wednesday and just stare at the screen?
not this week. i just reset the 5h and weekly limits for every paid user on mytabulon coding plan and kydanet.
clean slate. go ship something.
Opulent Byte retweeted
Kydanet too. Every paid plan just got its 5-hour and weekly limits reset, so your Agents start the rest of the week on a full tank.
Limits reset 🔄
We just reset the 5-hour and weekly usage limits on every paid MyTabulon Coding plan. Whatever you burned through this week, it's back.
Go build something.
platform.mytabulon.com/prici…
Opulent Byte retweeted
Limits reset 🔄
We just reset the 5-hour and weekly usage limits on every paid MyTabulon Coding plan. Whatever you burned through this week, it's back.
Go build something.
platform.mytabulon.com/prici…
the memory on kydanet is just amazing, things that happened months back, the agent remembers everything! every detail. a very powerful tool. we did magic on memory.
I am very impressed, memory is usually difficult to get right, but we did.
We've been quietly building something for a while and I finally get to show it.
It's called @Kydanet. You make an AI agent, and instead of living in a chat box it gets its own computer. A real Linux machine where it can open a browser, run code and keep files. It also gets its own email address, it can message people and post, and it has a little face that shows you what it's up to.
So you can tell it "find me the best noise cancelling headphones under $200" or "sort my inbox before I wake up" and walk away. It does the work on its own machine. You decide what it's allowed to do on its own and what it has to ask you about first.
The part I care about most is that it acts for you but never as you. It can't send email from your address or post pretending to be you.
It's free to start, no card needed, and it works everywhere, not just in the US. We're letting people in slowly while we add more machines, so there's a waitlist. Link's in the reply.
What's the first thing you'd hand off to it? follow @Kydanet for free invite codes too...
This is a really good model sir.
I'm really loving it, Atlas 1.5 is our best Atlas model yet. Feels like a Claude Opus 5.3 if it existed.
Opulent Byte retweeted
Over the past few months, Anthropic scientists have engaged in deep research partnerships with six organizations developing new medicines. We've also begun drug discovery efforts of our own.
In doing so, we’ve learned much more about how Claude can support the many stages of drug development.
In this blog post, we share some early results from these efforts, and describe how this work fits in with our wider ambitions in the life sciences.
anthropic.com/research/early…
Opulent Byte retweeted
GPT-6.1 Terra is live on MyTabulon.
OpenAI's mid-tier GPT-6.1 for everyday agentic coding and writing, at half the price of GPT-6.1 Sol.
Pro + Max chat, Coding Pro + Max.
$1 in / $5 out per 1M tokens
Try it: mytabulon.com/dashboard/chat
Pricing: platform.mytabulon.com/prici…
Opulent Byte retweeted
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-b…
This needs to be done, will help a ton fight against misinformation, and grok can make it automated and faster too. @elonmusk
It's the weekend make sure you use Maximo Atlas 1.5 Cyber Blue to run checks on the code you and your AI wrote, secure your codebase.
You get free $5 api credits to get started.
Opulent Byte retweeted
Kimi K3.1 Pro is live on MyTabulon.
Moonshot's successor to Kimi K3 is available in Pro and Max workspaces, and on Coding Pro & Coding Max.
Text, image, and video input. Low, High, or Max reasoning.
Try it at mytabulon.com
Coding Plan pricing: platform.mytabulon.com/prici…
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
how do you defend against ai that finds holes faster than anyone can patch them? you put the same ai on your side.
this week ai found php2shell, two critical bugs in php (the thing ~70% of the web runs on). it leaked two days early and sites went dark, from wikimedia to 32 uk government services. a few days before that, reuters reported one attacker used ai agents to break into 100+ company websites for about $50 a break-in.
attackers already have ai. most defenders don't. so we built maximo cyber:
🔵 cyber blue is for defenders and it's open to everyone, free plans included. we gave it one prompt and a little php shop. it worked out the shop was exposed to php2shell, patched it, then found and fixed 7 more holes. 6 min 40 s. 34 cents.
🔴 cyber red is for the people paid to break in, with permission: pentesters, red teams, researchers. it'll actually do the work, so we check who's asking. id checks, and a real person reviews every application.
both cost exactly what atlas 1.5 costs. the only model that beat cyber red on the benchmarks we compared costs 50x more.
go break your own stuff before someone else does 👇