25+ yr builder now shipping real projects with AI - Raw field reports, tool battles & rescues! #AriesShipLog
Sydney, Australia
Joined March 2025
- Tweets6.3K
- Following1.1K
- Followers855
- Likes8.5K
Pinned Tweet
Who would have thought that asking to connect would lead to meeting some really cool people on X.
Had some cool interactions with you all. I feel like this is what social media was supposed to be all along.
Thank you!
Last time today, but for always: "Happy to connect" ✌️
I felt this "wtf" bomb from @theo haha.
WTF?!?! 🤯💯
RE: Amp: "You can just self host Orbs now"
MicroVMs are all the rage now, but you should know that Amp was doing it way before it was cool 😎
To all the solo founders connected with me. This one's a good one for you to learn.
If you want to understand one of the peaks of AI powered automation, then you should learn about the infra that goes into setting something like Orbs up.
Reminder: the state of AI as we know it only began less than a year ago.
We're watching the future of Computer Science form in realtime.
Take it all in.
AriesTheCoder retweeted
DeepSeek V4.1 Flash 1017 is now available for your agents. It features a massive 1M token context window and smart reasoning out of the box.
developers.cloudflare.com/ch…
OpenAI knows that optimising for value is the long play.
Thanks guys. Still a fan btw - come back soon.
Once you guys get tired of "pacing the frontier", I'll be right here with my wad of cash and bag of coins 💸💰
RE: Hermes + Cua Spaces
I'm noticing a trend here.
First it was just agents,
then it was agents using browsers
then it was agents orchestrating
then it agents using browsers
then it's agents using your computer
then it's agents with their own sandboxes either locally or in the cloud.
RE: Takku - transcription app
Check out Takku by @MameliFilippo - It's a transcription app that writes notes directly into your Obsidian vault.
Will be trying it out im some of my meetings check out takku.app/
I've been using my transcription app for a few weeks now, at work and outside of it, and I think it's time to introduce it properly.
It's called Takku, and this is the site: takku.app
It has replaced Granola for me, and I'm not going back. It's lighter, it's faster, and I don't run into the limits of Granola's free plan. It also costs next to nothing: with GPT-6 Luna, Haiku 5.5 or DeepSeek, an hour-long meeting costs cents.
I optimized it for the Mac and didn't add any useless features. I want it to do one thing and do it well. The Unix principle.
The latest addition is Hooks: when a meeting is ready, Takku runs a script of yours. That's how it plugs into the rest of your system, for example by publishing the meeting to Notion.
A few colleagues and friends are already using it, and the feedback has been good so far.
RE: Overalls - AI First Platform by the Ralph loop guy.
My countryman Geoff Huntley building out his own ghost (or is it cat) in the machine (embedded model with MicroVM access)!
Gonna check out overalls.dev later today to find out more.
in other news; i’m ditching flutter.
going for rust core
then native mobile/desktop apps per platform for overalls.dev
Pretty happy with Claude Code's performance and uptime ATM.
@AnthropicAI got the server resources issue under control?
RE: DeepInfer is now Open Source!
@deepseek_ai JUST open sourced DeepInfer!
If you've been curious about how how AI inference engines and workloads are handled, check out the repo!
Github repo:
github.com/deepseek-ai/DeepI…
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.
🔗 github.com/deepseek-ai/DeepI…
1/4
I remember when I first set Win98 up at my house for my bro and I off a burned copy of Windows 98 CD from the new group of friends I had met.
It was only for a couple of years, but they were the best years. FFWD 25 years later we're creating homages to that magical decade.
oh yah... and @herdrdev is awesome.
⚖️They say CLAUDE is the best LEGAL AI.
The test: One AI to rewrite an existing agreement for a completely different business.
WHO WINS? 👇
0%ChatGPT
100%Claude
0%Gemini
0%Other (Comment)
1 votes • Final resultsRE: Minimax M3.1 Flash is out!
Frontier-level results at budget pricing.
5 Things to know:
- 73.7% DeepSWE 1.1
- Scored 82% at CyberGym
- $0.15 in, $0.60 out / 1M tokens
- Leads open weight models in computer use
- Multimodal!
MiniMax M3.1 Flash: Space Performance + Bunny Speed
🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.
💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed.
API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens.
Model weights will arrive on Oct 23.
API: platform.minimax.io
M Plan: minimax.io/m-plan
MiniMax Code for desktop: agent.minimax.io/download
So I've been playing around with plan ref tools - a persistent planning tool for agentic coding - more on this another time.
In doing so, I found that claude desktop now supports linux - amazing. If you're a linux user and have a claude sub, claude.com/download
ALT https://claude.com/download
Yet another one. We keep getting these reports on how great it is. So happy to see it.
Let's find a way to standardise te quality of these open weight models.
🚨 October Release (1017) DeepSeek V4.1-Flash
A different version of the DeepSeek V4.1 Flash model has been relesed and is actually going to be one of the first legitimate Open Weight contenders for frontier based off the benchmarks alone.
🚀 DeepSeek-V4.1-Flash-1017 is here!
Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.
🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.
1/4
I've been using the Sep flash model since it came out and it's already solid, thisis an improved version of that!
The model has been optimised for use with @DeepSeekHarness.
Disappointingly, they only provide Desktop for Mac and Windows only - with web only Linux support☹️
