AI Automation Eng • @AnthropicAI Model Safety Tester | Founder northwave.ai | Building in 🇨🇦

Joined September 2011
2,258 Photos and videos
topmass retweeted
Jun 13
This is going to have an opposite effect that the decels want. It's a huge open source AI accelerant, an accelerant for corporations, and enterprises Now it's a real race. You either make your own AI infrastructure or you don't have a seat at the table
97
203
2,555
91,093
Jun 12
how to rate limit resets on CLI ? @thsottiaux
2
53
Jun 12
from NPM to AUR so it begins
🚨 BREAKING: More than 400 Arch Linux User Repository packages have been compromised with infostealer malware and a rootkit. Attacker posed as a trusted maintainer and "adopted" orphaned packages. Arch maintainers are purging infected packages now. Audit your AUR installs.
67
Jun 12
Openai MUST respond before the 22nd and then (ideally) anthropic goes oh crap and keeps fable in subscription
Claude Fable 5 is in many ways the most powerful AI model ever trained. That shouldn't come as a surprise. It's the largest, most expensive and most recent model from a major AI lab. It won't be long before we're saying the same thing about another lab's model. What I have a hard time with is committing to Fable long-term. It's very expensive via API, and is only guaranteed to be available with a fixed-fee subscription for a couple weeks. It's like dating someone you know there's no future with. What's the point? All the major benefits come from the compounding over time. I realize that this is only the case for me as an individual consumer. The calculus for my clients is completely different. But I do think it's sad that the most capable AI technology is beginning to fade out of reach for the average consumer.
1
3
1,092
topmass retweeted
Jun 11
Introducing OpenQuick - Fully open source variant of shopifys Quick super configurable - easy to deploy and runs each quick page on a single durable object - thanks @CloudflareDev Check it out ⬇️
Everyone's talking about AI-generated HTML. But have you tried giving your sites a zero-config API for saving data, file storage, AI, websockets, etc? We did this at Shopify. Runs on a single VM that costs $200/month, and it's changed the way we work. We call it Quick 👇🧵
1
2
2
461
topmass retweeted
Replying to @rileybrown
Mf’ers aren’t even hello worlding anymore
2
2
30
1,950
Jun 11
Introducing OpenQuick - Fully open source variant of shopifys Quick super configurable - easy to deploy and runs each quick page on a single durable object - thanks @CloudflareDev Check it out ⬇️
Everyone's talking about AI-generated HTML. But have you tried giving your sites a zero-config API for saving data, file storage, AI, websockets, etc? We did this at Shopify. Runs on a single VM that costs $200/month, and it's changed the way we work. We call it Quick 👇🧵
1
2
2
461
Jun 11
github.com/topmass/openquick Open quick only requires a cloudflare account and wrangler and the install is simple out of the box it supports cloudflare auth landing page for quick drops or just use the CLI with "oquick ." inside a folder super fast built in wrangler ai too!
180
Jun 11
Just finished building Openquick works great! - Every site runs on a single durable object on @Cloudflare - for any assets above the 25mb limit they just get their own folder in the R2 bucket - CLI included, cloudflare auth included easy to deploy on free or paid !
77
Jun 11
yesterday fable 5 was speedy and fun today fable is slow and im not sure yet it is still whatchamacaliting
46
Jun 11
Vancouver BC founders what are you building ?
2
62
Jun 11
I have been working on goalforge - my answer to loop engineering and fable has made it excellent even with local models - It started when codex released "goals" the first iteration of loops. I've been working on this for a while - excited to share more!
48
fable 5 recommended using cloudflare email sending service on its own vs resend and its still in beta from just knowing that I lean toward using cloudflare in my stack that's pretty nice!
74
does anyone else feel like claude fable is way dumber now vs launch ?
1
1
53
lol claude opus 4.8 just decided to take a 3 minute break for no reason during a loop he was tired i guess
70
i have finally hit limits on pro 200 since no longer having 10x bonus @thsottiaux i am building a very cool codex based agent loop harness pls fix
1
89
have been working on something cool surrounding loops for a while already goalforge is a serious loop based codex orchestrator
53
if you don't have a local GPU farm with 10-15 webcams watching your facial expressions to tell the meta agent that infers what loops you would have wanted if its doing a good job or not you're ngmi
if you’re still writing loops that prompt coding agents you’re falling behind. you need to build a meta agent that infers what loops you would have wanted based on your vibe and then write those loops
74
pour one out for codex fast mode tonight GPT 5.5 party extra usage limits end on June 5th
1
127
literally never using tmux again or going back to anything else herdr is so good
1
149