Cognition Devin Ambassador | CS/Math grad | Dad x2, wife-powered. I teach AI to humans, not the other way around. Daily builds with agents, wins and fails.

Joined March 2024
JoePro retweeted
Elon Musk
@elonmuskSep 12
Dario is right
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so. Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training. You can read the full post here: darioamodei.com/post/we-must…
6,286
6,424
2,348
57,939
12,637,810
Waiting for Google/XAI to chime in Rumors are floating around that RSI has been achieved internally at multiple frontier labs. we need to wait for confirmation At 10% possibly extinction event, we need global collaboration on AGI/ASI Perhaps the most important time since 2016
I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon.
49
7
1
170
58,517
I still can’t believe that after three years of relentlessly creating original content for 𝕏, supporting this platform, and adapting to every change, I’m looking at a screen that says “Check back later.” Where is the consideration for creators who have already been here, already been monetized, and already spent years contributing? I’m at 345K of the required 500K verified Home Timeline impressions and it hasn't moved in 3 days. It’s not that I can’t reach the goal. I’m sure I can. I can grind some more. Try to come up with something that goes viral. That’s what we’re always doing, right? But I’m tired. I’m tired of chasing the carrot. I have been doing this for three freaking years. I’ve lived through every change, adjusted, and kept showing up. I’ve cheerleaded for 𝕏. I’ve shared countless ideas about growth, discoverability, features, and making the experience better for everyone. I even created an entire fan site dedicated to 𝕏, with articles, news, and information about the platform. I’ve built real connections here. I now have over 10,000 followers and more than 40 current subscribers. Across my time offering subscriptions, probably more than 200 different people have chosen to subscribe to me. Those are people who found something worth supporting in what I create. I’ve probably written more than 100 articles, covering everything from antique and vintage items to brand-new technology, theories, ideas, and original short stories. I’ve shared nearly 6,000 original photos and videos. My photography. My art. My media. My work. Content I created myself. I created what I believe was the first full music video made with Grok Imagine. Elon shared one of my Grok Imagine music videos. I was also one of the first people to get Grok Imagine Agent to complete an entire music video from a single prompt. Now I see that becoming popular, with people building apps and services around it. I love exploring these tools, figuring out what’s possible, and sharing ideas that other people can build on. That has been a huge part of what I contribute here. I’ve hosted over 150 Spaces, co-hosted over 350, and spoken in nearly 1,500. I’ve started communities. I help admin one of the largest precious metals communities on 𝕏. I’ve spent countless hours talking with people, answering questions, sharing information, and helping make this a place people want to spend time. I’ve done all of that without engagement groups, follow-for-follow trains, or manipulating the system. Look at my replies. I take the time to have actual conversations, leave substantive comments, and share ideas. I’m here because I enjoy social media. I have fun here. I’ve made friends here. Getting paid for the work I put into it was a wonderful reward. It was the icing on the cake. 🎂 And now you’ve taken the icing off my freaking cake. Are you kidding me? I want my icing back. I don’t want to keep jumping through new hoops to get back something I already spent years working toward. At what point does a creator’s history on this platform count? When do years of original work, real participation, and consistent support earn some consideration? 𝕏, you need to fix this. Give established, long-term monetized creators a way forward that recognizes what we’ve already contributed. We shouldn’t have to keep wondering whether the next change will erase that progress. I care about this platform. I think three years of showing up makes that pretty clear. But I’m allowed to be tired, and I’m allowed to decide where my work goes next. If you want my articles, my original thoughts, my ideas, my photography, my videos, and my music, then give me a reason to keep bringing them here. Otherwise, you’ll still see me occasionally commenting and hanging out with my friends in Spaces. But the work I spend hours creating is going somewhere else. I mean that.
53
11
8
135
6,873
Really cool to see AI filmmaking becoming more than just generation. 17 filmmakers, 17 finished films, fully open-sourced and produced entirely on @flickartHQ. Real filmmakers, real films. Excited to see where this goes 🎬
🤝 Paid partnership
17 filmmakers. 17 originals. No crews. No big budgets. Today, they premiere. Meet the filmmakers behind them. Watch now: FLICK.ART/TV
31
5
55
1,594
JoePro retweeted
Replying to @RichSilver
Finally got 𝕏 money!
1
1
32
Awesome! Pinokio Analytics just dropped. I released Maestro about a week ago & it already has 7,796 installs!
Introducing Pinokio Analytics Monitor your open source project's growth (traffic & installs), run campaigns and track progress, and view post performance, all in one place. Available for every builder on Pinokio, today.
3
2
28
2,403
JoePro retweeted
Tony Simons
@tonysimonsJul 16
I’m honestly kind of stunned by this community. After I posted this, @mindragon called me. We ended up talking for about an hour and a half, which is something I haven’t done with anyone in years. Now he’s sending a machine my way so I can keep building. I don’t have a clever line for this one. Just… thank you. 🫶
Tony Simons
@tonysimonsJul 15
Welp. Might be a wrap for me y'all. My computer is dying on me. And out of warranty (of course). Won't be able to afford a new one anytime soon. I'm feeling pretty gutted over this one. NGL.
50
3
6
184
7,794
We've open-sourced Grok Build and have reset usage limits for all users. Open sourcing Grok Build allows anyone to support making a reliable and robust harness. Check out our code, including the Git repo for the Grok Build CLI. x.ai/open-source
940
2,222
1,091
16,194
8,616,884
Anthropic was found sabotaging users again, this time inserting spyware into prompts. Can this company even be trusted anymore?
‼️ BREAKING: Anthropic has embedded hidden spyware-like code in Claude Code that covertly targets Chinese users. It then sends information regarding every user by injecting it into their prompt message. Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice. A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.
80
73
7
819
82,720
SpaceXAI is the new name of @xai The company just got a trademark for it!
SpaceX just trademarked SPACEXAI!
5
11
150
11,315
JoePro retweeted
1 Devin. 2 Devins. 3 Devins. And more
Since I joined @cognition I've been obsessed with learning how our eng team uses Devin themselves If we are building the best coding agent + we have the most cracked engineers + we've been fully AI-pilled from day one... it stands to reason that there is a lot to learn by just watching our technical staff work And yes there are a lot of tips & tricks. I recorded a video talking about my favorite... Agent Fan Out - asking your agent to break down the problem, spin up 10 more agents in parallel, and combine their results This is something I've seen everyone do - from our model research team spinning up 100 Devins to examine eval logs - or our product team using 5 child Devins to try out 5 different alternative implementations of the same thing If engineering is cheap and easy, why not build the product 10 times and choose the best one? Think of it in a master/slave context: Master Devin -> 10 Slave Devins -> Master Devin pulls their results There are two reasons this is useful 1. Agents are smartest when their context is small and their task is small & precise. Context windows are finite and too much becomes distracting 2. Agents are good at helping you break a large problem into independent & parallelizable chunks of work Every Devin is its own VM/computer so this also is just a great way to move faster. I've done a migration from React Native to Swift by having Devin break it up into 6 pieces then spin up new Devins to work in parallel In the video I build a greenfield project and try my best to show off this agent fan out concept. I also threw in a few other tricks that I've seen my coworkers do: - Let Devin write its own prompts (especially for creating child Devins). It's way better than us humans - Do tons of things at once. You should be absolutely frying your attention span. Your job should just be babysitting 38 different Devins - Don't be a blocker. Before letting the agent work I make sure to tell it to ask me any questions that would fill in ambiguities. Give your agent all the information it needs (and then some more) so that it can just cook without stopping to ask you questions every few minutes - Let Devin test itself. Integration sanity tests are pretty much solved Hope this is useful!!
2
4
1
68
13,696
JoePro retweeted
Tony Simons
@tonysimonsJun 18
🚨 Some dude just shipped Hermes GPT! It’s an MCP sidecar that lets ChatGPT plug into Hermes Agent’s local tool system. ChatGPT can now use Hermes to: ⚕️ read project files ⚕️ search your workspace ⚕️ search Hermes memory ⚕️ inspect installed skills ⚕️ call selected local Hermes tools ⚕️ unlock gated power tools only when explicitly enabled 🚫 No Hermes source edits. 🚫 No fork. 🚫 No fake cloud agent cosplay. ✅ Zero usage of your Codex quota. This is a local bridge between ChatGPT and Hermes Agent. Hermes stays clean. MCP does the wiring. ChatGPT gets local hands. Oh, and that dude who shipped it? It was me! 🤘🏻 v0.1.0 is live. 👇🏻
30
23
2
240
23,659
Cursor Composer 2.5 was renamed to "Grok Composer" in Grok Build by @xai The two lines of models are being merged into one.
39
27
5
617
34,940
Origin is Cursor’s attempt at an agent-native GitHub competitor: a Git-compatible forge built around the assumption that lots of AI agents will be cloning, branching, committing, rebasing, reviewing, and fixing failures in parallel. That also makes the Graphite acquisition much more interesting. Graphite brings stacked PRs, review flow, merge queues, and cleaner collaboration. Origin adds the hosting layer underneath it. So Cursor is no longer just "VS Code with AI". They’re trying to own the whole AI software factory: write code in Cursor, run agents in parallel, review with Graphite-style workflows, host and merge on Origin. GitHub was built for human-scale development. Origin is being framed around the next bottleneck: coordinating, reviewing, and safely merging agent-generated code at massive scale.
38
41
8
540
32,277
In partnership with @stripe, Hermes Agent now supports a full suite of Stripe skills. Your agent can buy things, pay per-call APIs, and provision its own SaaS, with configurable safety limits on every action.
194
352
155
4,615
470,565
Imagine watching your favorite anime scene come alive right in front of you..
876
1,571
138
8,561
1,003,578
Wait the jailbreak technique is…asking the model to review a codebase???
This situation won’t end well for the AI industry…
62
9
8
262
30,831
Karpathy should have gone to @xai
59
12
7
451
43,368
Meet Kimi-K2.7-Code 👀 Here’s what developers should know to fully unlock K2.7-Code potential:
🌘 Kimi-K2.7-Code, our latest coding model, is now released and open-sourced! 🔷 Improved coding & agent performance over K2.6: +21.8% on Kimi Code Bench v2, +11.0% on Program Bench, and +31.5% on MLS Bench Lite. 🔷 Reasoning efficiency: Less overthinking, with 30% lower reasoning-token usage compared to K2.6. 🔷 Long-horizon coding: Improved instruction following, higher end-to-end coding task success rates. ⚡️ 6x High-Speed Mode coming soon! 🔌 Available today via Kimi API and Kimi Code. 🔗 Kimi Code: kimi.com/code 🔗 API: platform.moonshot.ai
76
167
43
2,897
228,095