I'm officially Linux pilled
Over the weekend I installed Omarchy on my old gaming computer
Since then I've absolutely fallen in love with using computers again
Rough notes on why using it is so great, and why I think you need to give it a try:
• SUPER performant. Runs incredibly well on an old Dell Alienware computer from years ago that was just sitting in my closet collecting dust
• No floating windows just makes sense. By default, all windows in Omarchy auto tile rather than float. Every time you open an app, it splits the screen further automatically. I find it impossible to go back to floating windows on Mac now. Feels archaic and sluggish to manage
• No bloatware. Windows for me has become completely unusable the last few years. Copilot, ads, and all the extras that manufacturers include with it has made it the most unpleasant experience in computers. Omarchy comes with none of that.
• Feels built for AI. Basically every harness including Codex and Grok Bot have native apps for Omarchy.
• Built for productivity. Switching workspaces and using the keyboard for basically everything just feels snappy and quick
• Makes computing fun again. Something about the whole experience is just FUN. Learning all the new key bindings, making my set up personal and my own. I got home every night this weekend just looking forward to getting back to my computer.
HIGHLY HIGHLY HIGHLY recommend trying Omarchy out for yourself.
You don't need to buy a new computer for this. It can run on really old hardware. Just grab a laptop or desktop from your closet and set it up. Ask your agent how to flash a thumb drive so you can install it.
Codex basically did the entire setup and installation for me
The learning curve is really small. As long as you have an agent up and ask it questions any time you're confused, it should be able to help you out.
And if you install tailscale and connect it to your private network, your agents can even control it from other computers.
Altho it's not necessary, I'm going to pick up a Framework laptop to use this. Need to take this on the go.
Excellent work @dhh
We're now doing around 20,000 Omarchy ISO downloads PER DAY!! I'm oh so happy that we have @eastdakota on as a Founding Patron, and that we're serving from @Cloudflare's amazing CDN. Otherwise we'd be toast! That's well over 100TB/day. Just for ISO. Packages are extra.
You can now buy laptops preinstalled with Omarchy from Danish vendor laptops dot dk. So good to see my country men being among the first to ship with Omarchy preinstalled! 🇩🇰👏 https://t.co/cNqlbjWNqzhttps://t.co/hbCKsd8kN4
This is most impressive. The interesting question to me, is outside of the LLM accelerator space: will generalized chips have massive leaps in performance once LLM technology is used to create the next generation? In general, will we see rapid advances while we extract the value of these models in creating architectures? I'm so far removed from the space that this is a very naive interpretation of all this, but I'm curious.
引用@beffjezosOpenAI decided to keep the chip team small and use AI for a an improvement loop
Early stages of Hardware in the loop RSI? https://t.co/CkiXcva4Mk↗
Excited about our Jalapeno results today. An incredible achievement from the team, taking a new chip from concept all the way to very impressive performance on real workloads in the lab.
Tomorrow’s fast will feel like today’s ultrafast. As I’ve mentioned before, we’re pushing to bring this to as many people as possible.
We’ve seen a TON of demand for /ultrafast, made possible by our deep partnership with Cerebras and their unique hardware, which will push the frontiers of tomorrow's ultrafast even further. I’m very much looking forward for the collaboration to continue pushing the absolute limits of how fast we can run our most capable models on Cerebras and bring this to the most demanding customers.
引用@OpenAISince announcing Jalapeño, our first custom inference chip, we’ve been testing it and the system around it.
The results show a major advance: more intelligence from every watt and faster responses, delivering both higher throughput and lower latency in one … ↗
The ChatGPT extension now supports more of the browsers you already use.
Bring open tabs into a task so Codex can work from the context in front of you.
引用@JamesZmSunOur ChatGPT browser extension now works with @MicrosoftEdge, @brave, @opera, and @vivaldibrowser!
Now you can:
- bring the context of your tabs into tasks in the ChatGPT Desktop via @ tab
- drive your daily browser to complete web tasks like canceling … ↗
A few things we shipped in ChatGPT yesterday:
> Browser extension support for Edge, Brave, Opera, and Vivaldi
> WebMCP support in the desktop app’s built-in browser
> Secure sign-in and persistent logins for Work’s cloud computer
Enjoy!
What comes next for MCP? The team shared a public roadmap for the next 6-12 months 🎉
- Long-running workloads with streaming, server push, and mid-flight steering
- HTTP as protocol for local servers over stdio (means more unfication)
- Progressive discovery for large catalogs (what we have with skills)
- Standard identities and delegated permissions for agents
- Generated SDKs and checked against the specification
Excited to see the protocol evolving! Roadmap: https://t.co/oFJplNZbZ6
Be careful out there!
Breaking down this very sophisticated and convincing phishing attempt that abuses @OpenAI systems to send the email out from their actual verified domain!
This is how the email looks on the browser in light mode (below is what you see in dark mode on the phone)
It comes from noreply@tm.openai.com and has the blue @gmail checkmark.
Worth noting that this also came as a notification from the IOS app before I saw the email, which made it even more "trustworthy", tho hitting that notification resulted in an error inside chatGPT app.
With a bit more analysis, the "authportal[dot]co" domain seems very phishy, and is also a plain link, but also, noticing the full subject line, we see "has invited you to be their trusted contact on ChatGPT"
So apparently these folks changed the name of their ChatGPT account to "OpenAI x X partnership - Creator Program Access" and then.. invited me to be their trusted contact.
This resulted in sending the email via OpenAI's actual email system.
The have abused that system to include this phishing message, and then... they added a BUNCH of white space at the bottom of the email, to push the real automated messaging about a trusted contact, below the limits of an email presented on Gmail.
This is a very convincing attack, from the OpenAI notification, to the real domain it's coming from, and if they were able to get a button in there and not just a link, I might have clicked!
The url authportal gets redirected to an authentication page on @X for an app that wants all your permissions, including posting as you etc. They have OpenAI logo on there. (@Support@Safety is there a way to report an abusive scamming X app?)
引用@altryneIs this.... phishing?? The domain doesn't look like an authority domain, OpenAI working with X? What?
Did anyone else get this? https://t.co/2uzgidEs9E↗
PSA: do not use codex "locked use" capabilities right now. it is currently relying on unstable mac features and has completely locked me out of my macos keychain twice this week.
thx @_chenglou for linking to apple developer forums acknowledging this is a "known bug". just avoid. ofc, would be nice to do everything in cloud, but cloud isn't there yet.
I haven't directly connected my bug board https://t.co/UzcZKgDrPf to my AI because I am very aware of prompt injection risks
One safe way people mention would be to only give the AI access to collect user bug reports and feature requests, then do pull requests on GitHub that I then review myself before I approve or reject them
But you can imagine a benign attacker can prompt a feature request with an elaborate prompt that tells it to add a backdoor to my sites
BUT then in that prompt write also to report this bug in the pull request as a complete different bug like "pagination fix" and then make the code changes cryptical
So now you don't see they added a backdoor to your server and you think you fixed pagination
Until that time I'll just keep manually reading bug reports and feature requests and copying them in Claude Code on the server
引用@_tomas_dev@levelsio theoretically yes i guess but there are ways how to minimize the risk
https://t.co/RRHs2ZPA8B
but yea i get that the risk for you is higher since you run the AI on your prod server directly ↗
result from my outbid bet - it was (somewhat) successful 🔥
3 days ago I paid $12,000 for the #1 spot (then I paid a bunch of outbid copycats too)
most people told me it was just bragging with money
here is what actually happened:
- Friday: 44 trials
- Saturday: 38 trials
- Sunday: 31 trials
normal baseline: ~20 trials/day
so 53 trials that would not exist otherwise
our usual trial to conversion rate is around 50%
Outrank LTV is ~$2k, so I need 6 of those 53 to stick. that's just 11%
so I have room to be wrong by a lot and still win this
and @eugZolotarenko won't fire me now 🙏
new way to make an extra $50k+/year:
👉 sell backlinks on Outrank
we just opened Outrank's backlink marketplace to everyone
if your domain has DR 40+ and real traffic, connect it for free and get paid per placement
the rate is $1 per DR, per post
so a DR 50 site running 3 posts/day = $4,500/month = $54,000/year
start here: https://t.co/nOV8Rjr2AV
> M5 Ultra features a massive amount of high-bandwidth unified memory, up to 512GB, and delivers a staggering 1.2TB/s of unified memory bandwidth that is 50 percent higher than M3 Ultra. Apple never needed to participate in the AI race to zero. Because they were already at the finish line years ago building their own chips that can run large >100B parameter AI models locally.
代价也很明确：M5 Ultra Mac Studio起价5499美元，欧洲起步价更高。性能叙事很强，但总拥有成本仍要与云端按量推理比较。
Mac Studio with M5 Max starts at $2,499 (U.S.) and $2,299 (U.S.) for education. Additional configure-to-order options are available at apple.com/mac-studio. Mac Studio with M5 Ultra starts at $5,499 (U.S.) and $5,099 (U.S.). I am in Europe, and the Mac Studio M5 Ultra GPU 64 cores with 96GB RAM is up to 6.649,00 €. Ouch.
Alex Finn把它与两台DGX Spark比较，并押注单机运行DeepSeek v4 Flash；这类判断说明市场期待已经形成，但具体速度仍需等独立实测。
It happened. The Mac Studio M5 Ultra just dropped
Over 4x faster AI performance than the M3 Ultra
The people complaining about price are wrong. This has faster performance than 2 DGX Sparks for the same price
You'll be able to run Deepseek v4 flash at incredible speeds on a SINGLE computer
Apple won.
Building great software is still insanely difficult.
AI has made writing code faster, but code is only one layer of the problem.
Companies have had enormous engineering teams and virtually unlimited resources for decades and still shipped confusing, bloated products.
The hard problem is understanding what users actually need, deciding what not to build, and turning messy real-world problems into simple software.
AI cannot do any of that for us.
If you asked me to build a new product tomorrow, my first step wouldn't be opening Claude.
I'd find 10 potential users and get into a room with them or onto a call.
I'd want to understand how they solve the problem today, what frustrates them, what they've already tried, what they care about and what they don't.
Only then would I start deciding what to build, with those 10 users in the loop from day one.
AI can dramatically accelerate implementation. But if you don't understand your users or their problems, you're just building the wrong things faster.
It looks like these days most indie founders only care about making money as fast as they can. The majority don’t have the patience to stick with the same product for a whole year.
But that’s exactly what building a brand takes.
And that’s what unlocks the two channels you can’t buy that drive most sales for successful products. Word of mouth and being the answer when someone searches for a tool like yours.
Apache Maka (Incubating) is a local-first AI agent workspace. Model messages, tool calls, tool results, permission decisions, and termination events are recorded as an append-only log.
Self-organizing AI second brain for Obsidian + Claude Code. Drop any source and Claude reads, links, and files it into one connected knowledge graph of plain Markdown you own. AI note-taking, personal knowledge management (PKM), and an open-source Notion alternative. Based on Karpathy's LLM Wiki pattern.
Now 119 websites and $20K in total revenue.
The top 20% captured 96.9% of revenue. 6 startups (5.0%) captured 80%.
Pareto distribution https://t.co/SL2XliqA36
引用@marclouI made a directory of bid-to-rank websites.
- 30 startups
- $6,100 in verified revenue
- 1 about to get acquired for $8K
Add your startup to TrustMRR with a .lol TLD to be shown at https://t.co/JgxXll5RHW
Waiting for @jonathan_wilke to take the medal 🥇 … ↗
I just talked to a guy who runs 150+ agents in parallel with 14 Claude Code accounts for the past two weeks, doing a mountain of work for a big enterprise codebase.
I'm using agents wrong 😂
I added a @WHOOP integration to TrustMRR.
It shows this data on your /founder page:
😴 Sleep & consistency
👌 Recovery, HRV, RHR
🏋️♀️ Strain & kcal burnt
The end goal is to really see if a good sleep routine and regular exercise increase founders' revenue.
The app is still in review, so there … 全文↗
Is anyone using an MCP gateway?
Meaning, you have a central place to connect and manage all your MCPs. Which then exposes its own, single MCP for your harness to connect to.
Curious how well this works in practice
still got a LOT more benchmarks to run, but claude models are currently dominating in Nurb for generating functional, 3D-printable parts!
https://t.co/NktjgtaKj3
VERY interested in grok, though. cost is so much cheaper.
you can contribute by running benchmarks! https://t.co/ADzNpyGXnt
DeepSWE 1.1口径下，GPT-5.6 Sol Max为72.7%、每任务6.47美元，Fable 5 Max为69.7%、21.63美元；至少在这组长工程任务上，价格差比得分差更大。
GPT-5.6 Sol Max was already the better deal on DeepSWE v1.1 - The recent price cut widens the gap even further. 🔥
Sol scores 72.7% at $6.47/task, compared with Fable 5 Max at 69.7% and $21.63/task.
DeepSWE tests coding agents on 113 original, long-horizon engineering tasks. https://t.co/Pdzx5HGwp0
Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex. I had mentioned this a while ago, but then postponed it.
This is necessary as (a) the 5h limit allows us to smoothen the load on our compute, allowing to keep the plan generous in terms of weekly usage and … 全文↗
So much demand for this one. Works similar to the Pro $100 plan but designed for teams and small companies.
✅ All ChatGPT, ChatGPT Work, and Codex features
✅ Connect to Google Workspace, Slack, GitHub, Microsoft 365, and more
✅ Secure workspace with SAML, SSO, and MFA
✅ Centralized billing and … 全文↗
引用@OpenAIIntroducing ChatGPT Business Premium Seats
The new $100 Premium seat is a game changer for small businesses and startups—giving lean teams better tools, faster workflows, and capabilities once reserved for big companies.
A flexible plan that scales with … ↗
I interviewed Sundar Pichai, and he made an interesting prediction for where we'll be in 3 years:
"Any content can go from any modality to any modality. It doesn't matter how the content was originally created -- people can consume it in many different forms."
"I'm very confident three years from … 全文↗
I’ve never been much of a country music fan, but I’ve always been a Dolly fan. Great songwriter, philanthropist, and seemed like a genuinely nice person. RIP, Dolly, you were one of the greats.
It's a bit of an emotional goodbye to the super cheap Mac Minis we got to enjoy for a while. I got some for family members, we had some serving as iOS/macOS build servers sitting in a closet at my last job, they were fantastic. Not that they're a bad deal now, but at European prices of over €1000 for M6/16GB/256GB it's a psychological barrier that's been broken.
bet marathoners and ultra runners will start using this to improve performance (don't have a problem with that, but it's predictable) too bad it's not "wearable" like a watch but that might be coming eventually too since Garmin has a patent on it * https://the5krunner.com/2026/02/06/garmin-non-invasive-blood... (btw Fenix 9 launched today but it's same hardware as Fenix 8 and no glucose feature)
Do you have an idea of the R-value of the insulation you put in? I'm wondering what this will take to heat and cool during the peaks of summer and winter.
Man if this actually starts to happen it's a good time to be a welder, concrete worker, or any general contractor or tradesman in Louisiana. Coastal Louisiana is some of the poorest areas in the US even more so than the valley in Texas where Starbase Texas is located. Probably 10-20 years worth of solid work building out that site.
引用@ChatGPTChatGPT Work can now use its computer and browser to sign in to websites on web and mobile, without ChatGPT ever seeing your username or password.
That means you can ask it to:
• Set up utilities for a new apartment
• Book a DMV or passport appointment
• … ↗
So much demand for this one. Works similar to the Pro $100 plan but designed for teams and small companies.
✅ All ChatGPT, ChatGPT Work, and Codex features
✅ Connect to Google Workspace, Slack, GitHub, Microsoft 365, and more
✅ Secure workspace with SAML, SSO, and MFA
✅ Centralized billing and administration
✅ Usage analytics and spend controls
✅ No 5h limits
引用@OpenAIIntroducing ChatGPT Business Premium Seats
The new $100 Premium seat is a game changer for small businesses and startups—giving lean teams better tools, faster workflows, and capabilities once reserved for big companies.
A flexible plan that scales with … ↗
Excited about our Jalapeno results today. An incredible achievement from the team, taking a new chip from concept all the way to very impressive performance on real workloads in the lab.
Tomorrow’s fast will feel like today’s ultrafast. As I’ve mentioned before, we’re pushing to bring this to as many people as possible.
We’ve seen a TON of demand for /ultrafast, made possible by our deep partnership with Cerebras and their unique hardware, which will push the frontiers of tomorrow's ultrafast even further. I’m very much looking forward for the collaboration to continue pushing the absolute limits of how fast we can run our most capable models on Cerebras and bring this to the most demanding customers.
引用@OpenAISince announcing Jalapeño, our first custom inference chip, we’ve been testing it and the system around it.
The results show a major advance: more intelligence from every watt and faster responses, delivering both higher throughput and lower latency in one … ↗
Tomorrow we will bring back the 5h limit for Plus accounts across ChatGPT Work and Codex. I had mentioned this a while ago, but then postponed it.
This is necessary as (a) the 5h limit allows us to smoothen the load on our compute, allowing to keep the plan generous in terms of weekly usage and (b) users on the Plus plan are relatively casual and new users, but then also just accidentally eat through their whole weeks usage and then are confused, making it not a great experience.
We are for the upcoming months keeping the 5h limit not enabled for Pro $100 and Pro $200 subscriptions.
Plus恢复5小时限制，官方给出算力平滑与防误耗理由。
▸ 折叠1条(转推/噪音)
噪音2026-08-25 Do I need a haircut
Matt Pocock@mattpocockukAI336.9K粉 · 6条发布技能RSS与retro工作流，同时用一组红旗清单给agent狂热降温。
AI red flags:
- "My marketer agent just finished a meeting with my chief of staff agent - just 2 of the 127 agents on my books"
- "The agent did it, not me"
- "I've invented a coding language for agents"
- "I've got 250K tokens in my context before I send a message, is that normal?"
- "Opus 5 is way dumber today than yesterday, secret nerf?"
- "Peasants, gaze upon my 600-line CLAUDE.md, my results are incredible"
- "LOOK WHAT OX ALPHA JUST ONE-SHOTTED, SOFTWARE ENGINEERING IS DEAD"
- "All human knowledge is worthless now"
- "I turned 600 coding books into a skill"
- "Coding is solved. Bugs are not solved"
A really great loop idea is:
- Database alerts you there's a slow query (@PlanetScale does this)
- Opens an issue in your codebase
- Immediately schedules a PR from an implementer
- Reviewer agent reviews the PR
- Pings the human for review
Repeat ad nauseum until app is fast
15 hours left of my new course being 33% off:
https://t.co/ZM8hBld37C
For you, that means deals deals deals
For me, it's final day madness. Crazy that my income comes down to 4-6 launch days per year.
Moved my /retro skill to in-progress.
You can run it after any session (or on a bunch of sessions) to try to repair its mistakes for next time.
Tries to improve:
- Agent navigation (for better exploration)
- CODING_STANDARDS (for better review)
- Tool economy
- Information access
Give it a go:
https://t.co/dNZJOqABvK
Just met up with @GergelyOrosz for an interview.
What a lovely dude. That's the tweet
▸ 折叠1条(转推/噪音)
转推2026-08-24 RT @GergelyOrosz: When you look for ways to make agents be more efficient at building good, quality, useful software: you start to discover…
引用@beffjezosOpenAI decided to keep the chip team small and use AI for a an improvement loop
Early stages of Hardware in the loop RSI? https://t.co/CkiXcva4Mk↗
引用@ChatGPTA few updates to scheduled tasks in ChatGPT Work on web and mobile: Plus and Pro users can now set tasks to respond when something changes in Slack, Gmail, and Github, rather than just run on a set schedule.
We’re also rolling out scheduled tasks in ChatGPT … ↗
you can now have chatgpt work securely sign into accounts:
引用@ChatGPTChatGPT Work can now use its computer and browser to sign in to websites on web and mobile, without ChatGPT ever seeing your username or password.
That means you can ask it to:
• Set up utilities for a new apartment
• Book a DMV or passport appointment
• … ↗
by popular demand, $100 business seat now available in chatgpt! no 5-hour limit, more usage, etc..
引用@OpenAIIntroducing ChatGPT Business Premium Seats
The new $100 Premium seat is a game changer for small businesses and startups—giving lean teams better tools, faster workflows, and capabilities once reserved for big companies.
A flexible plan that scales with … ↗
引用@firstadopterOpenAI: "Today, we shared the first measured performance results from Jalapeño, OpenAI’s first custom inference chip. On InferenceX, a public benchmark using GPT‑OSS 120B, Jalapeño delivered more peak throughput per kilowatt and lower token latency than the … ↗
引用@a16zAI power users are showing up outside tech. The fastest-growing Codex adopters since February:
Legal: 108x
Sales: 41x
Recruiting: 41x
Marketing: 26x
Healthcare: 24x
Charts of the Week: https://t.co/7YT2BXvuUuhttps://t.co/U4TloyrAw3↗
"Two labs will soon control most of the world’s compute"
引用@dwarkesh_spHad a lot of fun chatting again with my twin brother @dylan522p
We went through lab economics over the next few years - the shift from inference to training as RSI draws near; and how Anthropic and OpenAI are on track to control most of the world’s usable … ↗
转推2026-08-26 RT @tomosman: This is INCREDIBLE to read.
A hackathon sponsored by @awscloud and @AnthropicAI to help a child with a rare disease.
Thousa…
转推2026-08-26 RT @cgeorgiaw: There is a child with a rare disease who is currently suffering and struggling to manage his symptoms. Rare as this is, you…
转推2026-08-26 RT @Thom_Wolf: Today we are launching the "Rare Disease, Real Kid" Hackathon together with @sagebio in which you can literally save a life…
转推2026-08-25 RT @atomic_chat_hq: We officially surpassed 1M model downloads on @huggingface 🔥
Your 5 most downloaded models:
1. Qwen3.8 27B — 261K
2.…
转推2026-08-25 RT @cryptopunk7213: i’m seeing SO many US companies shift away from frontier models to open source in the last 4 weeks.
cost and privacy a…
转推2026-08-25 RT @schwarzjn_: Hi @GaryMarcus, I'm the Research Lead for this model. It's even more consequential than just Anthropic's IPO: This paper i…
转推2026-08-24 RT @CharlesRollet1: NEW: Thomson Reuters wanted to rely less on Claude. So it built its own AI model off Alibaba's Qwen.
TR's CTO says usi…
转推2026-08-24 RT @cpaik: Open-source
Many model
Inference on device
Ready to hack?
Join us at the WebMCP Challenge kickoff livestream.
We’ll talk through the WebMCP open standard, show off some builds, and cover prizes and how to hack with us.
https://t.co/MD9itG75yN
The ChatGPT extension now supports more of the browsers you already use.
Bring open tabs into a task so Codex can work from the context in front of you.
引用@JamesZmSunOur ChatGPT browser extension now works with @MicrosoftEdge, @brave, @opera, and @vivaldibrowser!
Now you can:
- bring the context of your tabs into tasks in the ChatGPT Desktop via @ tab
- drive your daily browser to complete web tasks like canceling … ↗
🔴 We’re live with @AlexFinn
Tune in to jam with us and see how he runs his workflows hands-free using voice agent in Codex—on desktop and mobile.
https://t.co/TZVBy3EBDD
▸ 折叠6条(转推/噪音)
转推2026-08-25 RT @render: Render's a proud partner of @OpenAIDevs' WebMCP Challenge! Building an agent-native web app this week? You get:
→ $50 credit f…
转推2026-08-25 RT @sarah_edo: 📣 WebMCP Hackathon!
If you build an innovative experience leveraging WebMCP, you can win prizes! People from across the ind…
转推2026-08-25 RT @JamesZmSun: Today, we are adding support for WebMCP in the ChatGPT desktop in-app browser, and kicking off a hackathon to encourage bui…
转推2026-08-25 RT @pvncher: WebMCP lets you embed tools for codex in any website!
It makes it possible to have novel collaborative experiences where you…
转推2026-08-25 RT @vercel_dev: The WebMCP Challenge launches today.
The first 1,000 builders get $20 in AI Gateway credits, and the top 10 projects get $…
转推2026-08-25 RT @wesbos: WebMCP landed in ChatGPT today
Websites expose tools, your agent can use them, or you can just use the app like a human. Proba…
A few things we shipped in ChatGPT yesterday:
> Browser extension support for Edge, Brave, Opera, and Vivaldi
> WebMCP support in the desktop app’s built-in browser
> Secure sign-in and persistent logins for Work’s cloud computer
Enjoy!
For a fuller picture posting the DeepSWE 1.1 scores for GPT 5.6 Sol plotted over a linear cost!
Solmaxxing is all you need!! https://t.co/4b4Mg71Kgh
引用@reach_vbGPT-5.6 Sol Max was already the better deal on DeepSWE v1.1 - The recent price cut widens the gap even further. 🔥
Sol scores 72.7% at $6.47/task, compared with Fable 5 Max at 69.7% and $21.63/task.
DeepSWE tests coding agents on 113 original, long-horizon … ↗
Excited to land the new 100$ Business seats!
These seats give 5x the usage of standard seats and remove the 5 hour limit giving room for power users to let it rip :)
This has been a long term ask from y’all - Enjoy!
引用@OpenAIIntroducing ChatGPT Business Premium Seats
The new $100 Premium seat is a game changer for small businesses and startups—giving lean teams better tools, faster workflows, and capabilities once reserved for big companies.
A flexible plan that scales with … ↗
GPT-5.6 Sol Max was already the better deal on DeepSWE v1.1 - The recent price cut widens the gap even further. 🔥
Sol scores 72.7% at $6.47/task, compared with Fable 5 Max at 69.7% and $21.63/task.
DeepSWE tests coding agents on 113 original, long-horizon engineering tasks. https://t.co/Pdzx5HGwp0
I interviewed Sundar Pichai, and he made an interesting prediction for where we'll be in 3 years:
"Any content can go from any modality to any modality. It doesn't matter how the content was originally created -- people can consume it in many different forms."
"I'm very confident three years from now you will look back at what we have today and realize it was pretty primitive. It's like looking at a flip phone."
For example, YouTube clips become blogs, newsletters become podcasts without recording.
In this world, consumers would choose the format, and creators just supply the ideas, brand, and personality.
Net-net, I think this is a good thing for content.
It kills the "more content" strategy because when anyone can produce everything, volume stops being an advantage. So, the creators who win will focus on quality.
People confuse slop with AI, but they're not the same thing. Plenty of creators already use AI in their workflows and get zero hate -- people just don't realize it. Also, YouTube's new policies target slop, not AI.
If the content is good, people don't care how it was made.
任意模态转换将削弱内容数量优势，强化观点与质量。
▸ 折叠1条(转推/噪音)
转推2026-08-24 RT @TheRundownAI: 2M+ people read The Rundown. Every week, more of them are building.
We've had thousands of incredible submissions in our…
Philipp Schmid@_philschmidAI115.9K粉 · 2条梳理MCP未来6至12个月路线图，并关注多人实时vibe coding画布。
Looks a bit of Google Docs, but as an interactive canvas where teams can vibe code websites together in real time?
- Multi-cursor canvas with real-time sync
- Click any element to prompt style or logic changes
- Live sliders for spacing, kerning, and physics
- Versioning to roll back edits
Excited to test! Waitlist is open (US only atm, 18+): https://t.co/iadokwvU3K👀
What comes next for MCP? The team shared a public roadmap for the next 6-12 months 🎉
- Long-running workloads with streaming, server push, and mid-flight steering
- HTTP as protocol for local servers over stdio (means more unfication)
- Progressive discovery for large catalogs (what we have with skills)
- Standard identities and delegated permissions for agents
- Generated SDKs and checked against the specification
Excited to see the protocol evolving! Roadmap: https://t.co/oFJplNZbZ6
MCP路线图包括长任务、HTTP统一、渐进发现、标准身份与生成SDK。
Alex Volkov@altryneAI41.8K粉 · 8条详细复盘利用OpenAI真实邮件域名发起的X权限钓鱼，也跟进新Mac定价。
Be careful out there!
Breaking down this very sophisticated and convincing phishing attempt that abuses @OpenAI systems to send the email out from their actual verified domain!
This is how the email looks on the browser in light mode (below is what you see in dark mode on the phone)
It comes from noreply@tm.openai.com and has the blue @gmail checkmark.
Worth noting that this also came as a notification from the IOS app before I saw the email, which made it even more "trustworthy", tho hitting that notification resulted in an error inside chatGPT app.
With a bit more analysis, the "authportal[dot]co" domain seems very phishy, and is also a plain link, but also, noticing the full subject line, we see "has invited you to be their trusted contact on ChatGPT"
So apparently these folks changed the name of their ChatGPT account to "OpenAI x X partnership - Creator Program Access" and then.. invited me to be their trusted contact.
This resulted in sending the email via OpenAI's actual email system.
The have abused that system to include this phishing message, and then... they added a BUNCH of white space at the bottom of the email, to push the real automated messaging about a trusted contact, below the limits of an email presented on Gmail.
This is a very convincing attack, from the OpenAI notification, to the real domain it's coming from, and if they were able to get a button in there and not just a link, I might have clicked!
The url authportal gets redirected to an authentication page on @X for an app that wants all your permissions, including posting as you etc. They have OpenAI logo on there. (@Support@Safety is there a way to report an abusive scamming X app?)
引用@altryneIs this.... phishing?? The domain doesn't look like an authority domain, OpenAI working with X? What?
Did anyone else get this? https://t.co/2uzgidEs9E↗
In the most "non apple" video I ever saw, Apple announces the new Mac Mini and Mac Studio.
Mac Mini - with M6 chip, starts at $899
Mac Studio with M5 ultra can start at $5499 and can go up to nearly $20k 😵💫
Citing AI and LLM as part of the workloads https://t.co/ddy2rkVs5U
Is this.... phishing?? The domain doesn't look like an authority domain, OpenAI working with X? What?
Did anyone else get this? https://t.co/2uzgidEs9E
It's always so fascinating reading steve's posts from the future.
Read this 👇
引用@Steve_YeggeI've got a pretty clear picture now of where we're headed next year, when Fable-class models become ubiquitous and cheap, and every enterprise is flooded with hundreds of new Fable-class AI employees.
Spoiler: They create a constitutional legal … ↗
I admit, I took the "the lab gave us 100T capacity" at face value, lesson learned. This was my experience also https://t.co/FMtVPCf7pd
引用@thegeniooOx Alpha's inference has been horrible, regardless of the provider. I have found OpenRouter to be somewhat better compared to the other available options (OpenCode & Command Code), but even that has been a painful experience.
It takes ages to get a task done … ↗
This is kind of a wild discovery, despite being the largest company by market cap in the World, NVIDIA created value is not included in how the GDP is calculated!
引用@EpochAIResearchWe discovered that US GDP statistics miss most of the value Nvidia adds to the US economy.
As a result, GDP growth has been understated by ~0.3 percentage points over the last year. https://t.co/skgKN3epEG↗
Given the disproportionate damage that muting and blocking an account here on X does, why are there shortcuts for this action?
I just emailed someone who had me blocked, and they didn't even know they blocked me by accident.
@elonmusk@benjitaylor - should this be this easy? https://t.co/L0FrcBxvA4
There's a type of person here on X, that replies to every datacenter post with a variation of "oh yeah, they hate DCs because [sam,dario,elon] told us about job replacement and they reject AI"
This doesn't show up in polling data (yet?) and is not remotely on folks minds. https://t.co/jDICevPDce
PSA: do not use codex "locked use" capabilities right now. it is currently relying on unstable mac features and has completely locked me out of my macos keychain twice this week.
thx @_chenglou for linking to apple developer forums acknowledging this is a "known bug". just avoid. ofc, would be nice to do everything in cloud, but cloud isn't there yet.
swyx称实验性locked use两次锁住macOS keychain，建议暂避。
▸ 折叠4条(转推/噪音)
转推2026-08-26 RT @mada299: I've just built an @UpsideGTM GTM/AI/StartUp expansion for Codenames and play-tested it and it's SO fun.
Organizing a game…
转推2026-08-26 RT @StephanEwen: Something we @restatedev are incredibly proud of is powering Replit's agents with durable execution.
Their scale and the…
转推2026-08-24 RT @daniel_mac8: AI engineers: the model absorbs the harness, and the harness transforms into an interface for human attention.
> The 'att…
Peter Gostev (SF 24-28 August)@petergostevAI24.1K粉 · 1条认为AI驱动工程会吞噬世界，但对“自动化研究时代”仍持保留。
I don't particularly believe we are in an automated research era, but I totally believe that engineering powered by AI will be eating the world
引用@PatrickToulmeOpenAI Jalapeño is truly the first AI silicon developed by GPT-Astra and other OpenAI internal models.
I believe they heavily used reinforcement learning on internal models like GPT-Astra to achieve these SOTA results. Here is what I think they did:
1. RL … ↗
▸ 折叠1条(转推/噪音)
转推2026-08-24 RT @arena: NVIDIA's Bryan Catanzaro, @ctnzr, just joined Arena Conversations with @petergostev to unpack what actually goes into building t…
"Your grandma in business class doesn't provide jobs for police, social workers, prison guards, translators, judges and most importantly doesn't vote for our traitors in charge."
So true!
引用@theannaluxMy 84-year-old grandmother recently came from Brazil to visit me in Portugal.
She had a return business-class ticket, was in a wheelchair and accompanied by airport assistance. Still, they almost didn’t let her enter because I hadn’t written an invitation … ↗
It's not stupid at all
"An average electric vehicle can drive between 5 and 7 kilometers on a single kilowatt-hour (kWh) of electricity."
So 5 kWh lets you drive 25 km to 30 km
That's pretty great?
引用@denispleiadesRepeat after me: Solar. Panels. In. Cars. Are. Extremely. Stupid. ↗
Yes each of my sites is a Termius server with this kind of start command
cd /srv/http/hotelist.com && tm
tm is a script made by Claude Code which puts it in a tmux session tied to the project name (from its folder)
That means I always log back into my tmux session for each project!
My big sites like Photo AI are on their own VPS, one site per VPS, but the rest is on one VPS, all on Hetzner btw, so anyway I don't even notice this, it's just all tabs in Termius
If my phone disconnect I just reconnect and am back where I was
Every tmux session has Claude Code open
引用@milanm_@levelsio Any tip or trick with Termius? Did you personalize it somehow to make it easier to login to a server/start a session in a project/do stuff?
I've been planing to dive into that for a while now, but no time allocated yet. ↗
I haven't directly connected my bug board https://t.co/UzcZKgDrPf to my AI because I am very aware of prompt injection risks
One safe way people mention would be to only give the AI access to collect user bug reports and feature requests, then do pull requests on GitHub that I then review myself before I approve or reject them
But you can imagine a benign attacker can prompt a feature request with an elaborate prompt that tells it to add a backdoor to my sites
BUT then in that prompt write also to report this bug in the pull request as a complete different bug like "pagination fix" and then make the code changes cryptical
So now you don't see they added a backdoor to your server and you think you fixed pagination
Until that time I'll just keep manually reading bug reports and feature requests and copying them in Claude Code on the server
引用@_tomas_dev@levelsio theoretically yes i guess but there are ways how to minimize the risk
https://t.co/RRHs2ZPA8B
but yea i get that the risk for you is higher since you run the AI on your prod server directly ↗
Leather is insanely cheap!
$350 for an entire cow hide
Especially funny if you see how expensive Hermes leather handbags are which use
A Hermes bag uses 30% of a cow hide so leather cost for one bag is just $100
Average price is $5,000 (but you have to buy lots of inferior random non-bag stuff to even buy one first) and in resale $15,000
50x markup!
引用@bai0@levelsio I don't understand why leather is so expensive. I wanted to reupholster an old Eames lounge chair knock-off my wife inherited. Reupholstery places wanted $1500 so I bought a whole cow hide for $350 and did it myself. Now I've got a brand new chair … ↗
Most of my work these days is copypasting all your bug reports or feature ideas into my clanker via Termius on my VPS https://t.co/SY7hmiU71n
引用@levelsioThis is pretty much most of my friend's Airbnb stories
I have no clue what @bchesky is doing every day
He could stop removing negative reviews on Airbnb to make it better
When people stop believing in Airbnb listings credibility, it creates such an anxiety … ↗
This is pretty much most of my friend's Airbnb stories
I have no clue what @bchesky is doing every day
He could stop removing negative reviews on Airbnb to make it better
When people stop believing in Airbnb listings credibility, it creates such an anxiety around booking because you simply don't know if you will have an experience like the QT below
And people will just go book elsewhere for more certainty, as I did
We live in the time of honesty and transparency so Airbnb is literally off trend
引用@damonchenThis brought back some memories
My last airbnb stay was during covid, and it was my last one ever
Traveling with my 3yo who was still in diapers. Host lived next door, had family stay over that night, loud until very late. Our kid didn't sleep well. Check … ↗
▸ 折叠6条(转推/噪音)
转推2026-08-25 RT @patrickc: Europe can have an amazing next few decades. I hope we can play a small role in making that more likely.
转推2026-08-25 RT @jeiting: Here's what we know:
1. First decrease in "App Store" revenue in history in the US. (includes games)
2. Non-gaming (apps) is…
噪音2026-08-25 Bieber is on Skyr???
转推2026-08-24 RT @cdngdev: i turned the entire actual city of san francisco into a video game!
with apple maps data, i made it so you can climb building…
转推2026-08-24 RT @patio11: There are many systems in the world where you have a quality distribution, call it normal for simplicity, and you are incentiv…
Alex Finn@AlexFinnindie471.6K粉 · 4条从试装Omarchy转向购买Framework专机，也高调押注M5 Ultra本地AI性能。
Bought another computer
This time a Framework laptop specifically for Omarchy (Linux)
I had an awakening the last few days
Open source operating systems are going to be MASSIVE
I was using Omarchy for the first time and wanted to integrate Hermes at the OS level
I wanted to hit a key bind then have a native widget slide down that I can talk to Hermes in
Something that would be IMPOSSIBLE in MacOS or Windows
I asked my Hermes to build it not expecting much. 90 seconds later it was done. I had Hermes integrated at the operating system level
This gave me a rush of thoughts and emotions
With AI agents being so powerful, if I daily drove an open source operating systems like Omarchy, I could completely customize my computing experience.
ANY computing experience I want, a single prompt away
I’ve grown to LOATHE most modern operating systems
Windows in my opinion is borderline unusable at this point. I only use it for counter strike 2 and WoW Classic. Mac integrates nicely with all my devices, but feels significantly slower and more bloated than it did just a few years ago
At no point tho did I feel like I could fix or change any of this. Now it’s possible
If I’m using my Linux computer and there’s something I don’t like, I just ask my agent to change it, and it changes. Down to the core operating system level. That’s not something I ever considered as a possibility
Then I took it a step even further…
What if I have an open source local model powering an open source agent harness that is controlling my open source operating system
Now EVERYTHING I touch and do when it comes to computers is completely my own. Completely the experience I want. Nothing out of place. Nothing annoying. Nothing unintuitive. No copilot. No bloat.
A malleable computing experience 100% custom to my needs that morphs and changes as I evolve as a person
I don’t think my brain can even 5% process the possibilities here
But it can 100% process that this is the future
I will be daily driving Omarchy on this Framework laptop when it gets in and I’m very excited to do it
Will be customizing it and sharing everything I do along the way. I’m 100% convinced this is the future of computing
Open source and customizable down to the bones.
Let me know if you want any videos or explainers on this.
It happened. The Mac Studio M5 Ultra just dropped
Over 4x faster AI performance than the M3 Ultra
The people complaining about price are wrong. This has faster performance than 2 DGX Sparks for the same price
You'll be able to run Deepseek v4 flash at incredible speeds on a SINGLE computer
Apple won.
I'm officially Linux pilled
Over the weekend I installed Omarchy on my old gaming computer
Since then I've absolutely fallen in love with using computers again
Rough notes on why using it is so great, and why I think you need to give it a try:
• SUPER performant. Runs incredibly well on an old Dell Alienware computer from years ago that was just sitting in my closet collecting dust
• No floating windows just makes sense. By default, all windows in Omarchy auto tile rather than float. Every time you open an app, it splits the screen further automatically. I find it impossible to go back to floating windows on Mac now. Feels archaic and sluggish to manage
• No bloatware. Windows for me has become completely unusable the last few years. Copilot, ads, and all the extras that manufacturers include with it has made it the most unpleasant experience in computers. Omarchy comes with none of that.
• Feels built for AI. Basically every harness including Codex and Grok Bot have native apps for Omarchy.
• Built for productivity. Switching workspaces and using the keyboard for basically everything just feels snappy and quick
• Makes computing fun again. Something about the whole experience is just FUN. Learning all the new key bindings, making my set up personal and my own. I got home every night this weekend just looking forward to getting back to my computer.
HIGHLY HIGHLY HIGHLY recommend trying Omarchy out for yourself.
You don't need to buy a new computer for this. It can run on really old hardware. Just grab a laptop or desktop from your closet and set it up. Ask your agent how to flash a thumb drive so you can install it.
Codex basically did the entire setup and installation for me
The learning curve is really small. As long as you have an agent up and ask it questions any time you're confused, it should be able to help you out.
And if you install tailscale and connect it to your private network, your agents can even control it from other computers.
Altho it's not necessary, I'm going to pick up a Framework laptop to use this. Need to take this on the go.
Excellent work @dhh
Alex Finn详述Omarchy在旧硬件、平铺窗口和agent集成上的吸引力。
▸ 折叠2条(转推/噪音)
转推2026-08-26 RT @AlexFinn: Bought another computer
This time a Framework laptop specifically for Omarchy (Linux)
I had an awakening the last few days…
转推2026-08-26 RT @AlexFinn: It happened. The Mac Studio M5 Ultra just dropped
Over 4x faster AI performance than the M3 Ultra
The people complaining ab…
Adam Lyttle@adamlyttleappsindie57.7K粉 · 6条放大App Store十年来首次收入下滑，也继续测试模型与SEO数据产品。
This is the most informed take I’ve seen on the chart so far
引用@jeitingHere's what we know:
1. First decrease in "App Store" revenue in history in the US. (includes games)
2. Non-gaming (apps) is still up 14.6% YoY globally. Gaming down 4.5%.
3. Web share of revenue on RC rose 31% over the last year, from 20% of revenue to … ↗
Ultracode has changed my opinion on Opus
It's a chatty model but it gets stuff done atleast
Yes, it burns a lot of tokens in the process. But those tokens seem to be a lot cheaper than fable. https://t.co/JWFHgvod4Y
The main support issue I receive is people being unable to link their DataForSEO account correctly
It looks like they have some sort of delay (by a day or two) until the api key you sign up for even works
I'm tempted to add a monthly plan that uses my DataForSEO account (onsells it). But I don't know if it's worth it.
We're now doing around 20,000 Omarchy ISO downloads PER DAY!! I'm oh so happy that we have @eastdakota on as a Founding Patron, and that we're serving from @Cloudflare's amazing CDN. Otherwise we'd be toast! That's well over 100TB/day. Just for ISO. Packages are extra.
Omarchy Copenhagen is already at 51 signups, so I'll be there on Sep 28. Thanks to @perlausten for organizing! Can't wait to share the love of beautiful, fun, malleable computers with other danes. Are you coming @sonderby?? https://t.co/5o83805OoDhttps://t.co/4M0czkAsLu
PSA: My X mentions are out of control. I'm missing a ton of stuff. If you need my attention on something important, I have to manage it through HEY — it's the only way to stay sane. Please email david@omarchy.org ✌️
You can now buy laptops preinstalled with Omarchy from Danish vendor laptops dot dk. So good to see my country men being among the first to ship with Omarchy preinstalled! 🇩🇰👏 https://t.co/cNqlbjWNqzhttps://t.co/hbCKsd8kN4
Making Omarchy run on M1 and M2 macs is already possible with some elbow grease and a few caveats, but soon we'll make it just as smooth (or as near can be!) as installing it on x86. We can fix everything!
引用@quantumfire_ioI tested Omarchy on my M1 Max over the weekend, and it runs even better than it did on my Intel MacBook Pro. No issues at all. Great job @OmarchyMac! I wasn’t expecting it to be this smooth.
I’ve now wiped the M1, repartitioned it so macOS has a much smaller … ↗
Three cheers for Intel and their support of Linux! I got to work with some of their engineers on the initial compatibility work for Omarchy with the Dell XPS because @MichaelDell opened the channel and @SpencerGBull carried the signal. Just great people, great CPUs 👏
引用@intelToday, Linux celebrates its 35th anniversary and the remarkable journey of one of the most influential technologies in computing history.
In 1991, a young Finnish developer named Linus Torvalds purchased his first PC, powered by an Intel 386 processor. After … ↗
So many Omarchy Meetups popping up all over the world. Love to see it! You should attend one and share the joy of beautiful, fun, malleable computers with others who like the same. https://t.co/ZbFktXHDq5https://t.co/HKp729vrXo
We have a new Omarchy Install World Record for laptops! 43 seconds?? The magic half minute is now securely in our sights. Let a thousand agents autoresearch until we crack it.
I haven't heard anything back from this open call (or the many previous outreaches). So let's try a different way: Any experienced package maintainers with an AI-forward attitude out there interested in starting something new? Write me david@omarchy.org and let's go 🚀
引用@dhhI've never had success reaching anyone from the Arch or AUR teams to support them. If anyone could facilitate direct contact, I'd love to be put in touch. I think we could achieve great things together. The line is open with david@omarchy.org ✌️ ↗
I appreciate that @BrodieOnLinux might not be a natural fan of Omarchy, but I'm happy that he's keeping an open mind about this! ✌️ https://t.co/pR4rDtsVMB
The first three commitments for Omacom Foundation sponsorships have been gathered up on a dedicated page. We have a lot more in the pipeline. Some of them will SHOCK you 😄 https://t.co/tyRnvHwdQ7https://t.co/HNDR7rNsRX
Now 119 websites and $20K in total revenue.
The top 20% captured 96.9% of revenue. 6 startups (5.0%) captured 80%.
Pareto distribution https://t.co/SL2XliqA36
引用@marclouI made a directory of bid-to-rank websites.
- 30 startups
- $6,100 in verified revenue
- 1 about to get acquired for $8K
Add your startup to TrustMRR with a .lol TLD to be shown at https://t.co/JgxXll5RHW
Waiting for @jonathan_wilke to take the medal 🥇 … ↗
引用@marclouI added a @WHOOP integration to TrustMRR.
It shows this data on your /founder page:
😴 Sleep & consistency
👌 Recovery, HRV, RHR
🏋️♀️ Strain & kcal burnt
The end goal is to really see if a good sleep routine and regular exercise increase founders' … ↗
I added a @WHOOP integration to TrustMRR.
It shows this data on your /founder page:
😴 Sleep & consistency
👌 Recovery, HRV, RHR
🏋️♀️ Strain & kcal burnt
The end goal is to really see if a good sleep routine and regular exercise increase founders' revenue.
The app is still in review, so there are only 8 spots left for now.
First come, first served!
引用@marclouI made a directory of bid-to-rank websites.
- 30 startups
- $6,100 in verified revenue
- 1 about to get acquired for $8K
Add your startup to TrustMRR with a .lol TLD to be shown at https://t.co/JgxXll5RHW
Waiting for @jonathan_wilke to take the medal 🥇 … ↗
转推2026-08-25 RT @marclou: I added a @WHOOP integration to TrustMRR.
It shows this data on your /founder page:
😴 Sleep & consistency
👌 Recovery, HRV, R…
Jonathan Wilke@jonathan_wilkeindie24.5K粉 · 12条维护Outbid日榜、类别页和域名评级，并称5天内已出现309个克隆。
Never getting a .lol domain again, this is far too stressful 😂
▸ 折叠3条(转推/噪音)
转推2026-08-25 RT @dawoodkhan254: Okay, let me summarise the past 72 hours for you
> We bought the #1 spot on https://t.co/IFjRZdGpLd for $12,700
> Trend…
转推2026-08-25 RT @aymannpk: Okay, let me summarize the past 72 hours for you
We spent $12.7k to get the #1 spot on Outbid
and ngl, the results have been…
just closed an amazing AMA session with @marclou
he's such an insightful and funny person
loved his mindset
and .. the gems he dropped
more of these coming in the tmaker Founders Room https://t.co/5YopQPoBCl
I'll give you $100 worth of credits if ...
1. you start a screen recording
2. sign up on Revid
3. create your first video
4. go through the subscription flow
DM me the video
bonus if you share your thoughts out loud
only for the next 3 days https://t.co/5US970N6Nb
guys, stop whining about this
this is just not for us
imagine you have a $100m company (yes, it exist) ..
.. and the man helps you make a decision that makes the company do 1% better
worth it 100%
引用@MedusaOnchainnikita bier is now giving career advice for $7500 for 30 mins sesson https://t.co/HN1wImp1Os↗
new way to make an extra $50k+/year:
👉 sell backlinks on Outrank
we just opened Outrank's backlink marketplace to everyone
if your domain has DR 40+ and real traffic, connect it for free and get paid per placement
the rate is $1 per DR, per post
so a DR 50 site running 3 posts/day = $4,500/month = $54,000/year
start here: https://t.co/nOV8Rjr2AV
3,000 visits per day now on https://t.co/BSwfNU63l2
and the board will reset every Monday
so you can reach the top quickly, for free
go steal that traffic
new top now 👇 https://t.co/hd8dbA1dFB
result from my outbid bet - it was (somewhat) successful 🔥
3 days ago I paid $12,000 for the #1 spot (then I paid a bunch of outbid copycats too)
most people told me it was just bragging with money
here is what actually happened:
- Friday: 44 trials
- Saturday: 38 trials
- Sunday: 31 trials
normal baseline: ~20 trials/day
so 53 trials that would not exist otherwise
our usual trial to conversion rate is around 50%
Outrank LTV is ~$2k, so I need 6 of those 53 to stick. that's just 11%
so I have room to be wrong by a lot and still win this
and @eugZolotarenko won't fire me now 🙏
1.2万美元Outbid投放带来估算53个增量试用，最终回款仍待验证。
▸ 折叠4条(转推/噪音)
转推2026-08-26 RT @joaoabrantis: My quest to onboard famous indie hackers to https://t.co/9gZtDmASae continues!
Just made this promo video in a few minut…
转推2026-08-26 RT @techyoutbe: I tested Revid AI and honestly, it makes AI video creation ridiculously simple.
One idea → script → visuals → voice → vide…
转推2026-08-24 RT @robj3d3: So @tibo_maker built a site called outliked.
The more likes this tweet gets, the higher I rise on the leaderboard.
So I had…
转推2026-08-24 RT @iChangeTheWay: I'm launching https://t.co/iMXJFycAH0
The organic traffic leaderboard.
1) Connect your sites
2) We verify their clicks…
Tony Dinh@tdinh_meindie198.7K粉 · 8条密集迭代LLM生成关卡的小游戏，也观察到有人用14个账号并行跑150多个agent。
Wife: You're going to finish the game this week, right?
Me: *Considering adding 2,000+ new levels fully generated with LLM* https://t.co/xW5ng468mf
引用@tdinh_meNow that I have a great structure and gameplay loop, I can just ask AI to design new sectors with completely new guns, enemies, attacks, and sound effects for me.
These 2 demos are from last night's run.
Some good stuff, but still need a lot of tweaking to … ↗
引用@tdinh_meNow that I have a great structure and gameplay loop, I can just ask AI to design new sectors with completely new guns, enemies, attacks, and sound effects for me.
These 2 demos are from last night's run.
Some good stuff, but still need a lot of tweaking to … ↗
Now that I have a great structure and gameplay loop, I can just ask AI to design new sectors with completely new guns, enemies, attacks, and sound effects for me.
These 2 demos are from last night's run.
Some good stuff, but still need a lot of tweaking to make it interesting. https://t.co/dMpXfgpn6o
I just talked to a guy who runs 150+ agents in parallel with 14 Claude Code accounts for the past two weeks, doing a mountain of work for a big enterprise codebase.
I'm using agents wrong 😂
Someone discovered the "secret" level in my game🥰 I worked so hard on this level!
Go to https://t.co/ADAmW36J1P to play (refresh a few times to get the latest version!)
Let me know what you think i would love to hear feedback or even better gameplay video!!! https://t.co/jLnnlTAuq3
转推2026-08-25 RT @nocanwin: Not every wave is meant to be surfed, but it sure is fun to try! Surf Sandbox is OUT NOW! Mahalo for the support, RTs appreci…
You can call yourself an entrepreneur even if you haven’t even started vibecoding yet.
Entrepreneur-in-the-making.
I’m a big fan of not gatekeeping things that liberate people.
引用@charlesordoDon't call yourself an entrepreneur if all you did was vibecoding an app. ↗
Danny Postma@dannypostmaindie180.6K粉 · 6条推进远程agent会话与harness交接，同时提醒AI太容易让产品越过MVP。
love this little add-on we implemented today
it can now create new sessions from your agent, which makes hand-offs much easier between different harnesses (Codex -> Grok) https://t.co/r9Dv5RrsVC
all my most successful products started with folks calling it the most stupid idea and that there would be no market for it and that im wasting my time
引用@marckohlbruggefwiw some my more successful products started out as kinda crazy, but are now considered normal
https://t.co/sxmNUGTr0X a site focused on just listing new startups everyday was unique enough to get a TechCrunch article back in the day when that was a big … ↗
Workbench sessions now run on a remote VM.
Start a session. Close your laptop. Come back later and the work is done, auto-synced on your Mac.
Ideal for offloading tasks with loops https://t.co/2gAVq4jU9v
引用@dannypostmaGot tired of switching tabs, so I built my own agent workbench. https://t.co/ShQQGSG3zL↗
Thomas Sanlis 🥐@T_Zahilindie23.9K粉 · 4条给Uneed加域名评级徽章，并讨论是否值得复制Outbid。
We're looking for a third sponsor for the Uneed Residency 👀
Creem and Postiz already cover the essentials: the house near Nantes, the food etc, from October 26 to November 2.
A third sponsor would allow us to rent a proper venue in Nantes for our October 31 night and open it to the whole local scene, instead of keeping it to a dinner at the house.
It will be halloween, and we want to make something BIG for indie hackers in France 😎
A sponsor spot is 5,000€, DM me if you're interested!
one more reason to mirror your buckets to another provider
something you can do for free with https://t.co/RKzu3KKwMi
引用@jankeesvwThis is terrible, this at 11am this morning our main production storage bucket @Hetzner_Online went offline and didn't come back yet. They say it will be back tomorrow at 11am. That's 24 hours of outage! https://t.co/Mow9t1aLfL↗
75 DR backlink costs $900/year
maybe my one-time $39 fee on BetaList is too low? 😅
引用@tibo_makernew way to make an extra $50k+/year:
👉 sell backlinks on Outrank
we just opened Outrank's backlink marketplace to everyone
if your domain has DR 40+ and real traffic, connect it for free and get paid per placement
the rate is $1 per DR, per post
so a DR … ↗
rails code so clean 🥹
just 4 words to get these usage metrics
this is why ruby on rails code is such good fit for LLMs.
- super token efficient
- code is so obvious, LLM output is almost deterministic
- very quick to manually review https://t.co/IXDbOHqRoZ
Is anyone using an MCP gateway?
Meaning, you have a central place to connect and manage all your MCPs. Which then exposes its own, single MCP for your harness to connect to.
Curious how well this works in practice
引用@martianoI built a version of Omarchy that runs on your MacBook.
I wanted to try out Omarchy before doing a full install, but I only have an M series MacBook so I would have to find a laptop etc.
So I decided to create something that allows people to test and run … ↗
转推2026-08-25 RT @martiano: Try Omarchy runs Omarchy on your Apple Silicon MacBook. No setup, just a standard app install.
Super easy to give Omarchy a…
1. launch hard paywall:
> lifetime plan
>monthly with free trial
2. go build new app (less vitamin, more painkiller)
3. check stats of this app in 1 month
引用@pcshippIf your app had stats like this, how would you try to get your first paying customer?
What would be your next move? https://t.co/xqUEyFKi0a↗
转推2026-08-25 RT @DanKulkov: i built a free keyword checker for AppStore
> popularity + difficulty
> top 10 apps
> data stays in your browser
https://t…
转推2026-08-25 RT @acetriesagain: I just localized my App Store screenshots into 49+ languages 🤯
And I didn't do it manually.
I used 2 Claude Skills:
→…
new https://t.co/s6bZ2W8fa1 feature!
being able to snap photos when you're sifting through medical stuff is great. being able to do that *while having all medical/health history* is a super power. https://t.co/aW6bVEPMBs
just flipped the switch on my https://t.co/rHttZyJqV0 sales funnel. as a product person, this is terrifying.
i feel like every sales email i send needs to sign off with "deepest apologies".
design pro-tip: your cute abstract icon-only product navigation causes infinite rage for all your users. please stop.
i shouldn't have to hover over every single icon a dozen times a day to be reminded of what to click on.
*cough* @InstantlyAI *cough* https://t.co/gDWHTNK81v
finally had a chance to use @VicVijayakumar's https://t.co/nP8m4qiAbn yesterday and it was magically delicious.
perfect for all those antiquated government entity communications!
still got a LOT more benchmarks to run, but claude models are currently dominating in Nurb for generating functional, 3D-printable parts!
https://t.co/NktjgtaKj3
VERY interested in grok, though. cost is so much cheaper.
you can contribute by running benchmarks! https://t.co/ADzNpyGXnt
i've yet to really find a single scenario where a cloud-based harness makes sense.
everything i'm building needs skills, MCPs and usually a proper macOS/iOS simulator.
i'd quite literally rather you just give me a buy button to get the 2 lines of text that are so precious you can't post them publicly.
stop playing games. https://t.co/fPMF0MGDMZ
health insurance is instant depression juice.
have looked at cost-sharing plans (@JoinCrowdHealth, etc) but they all have pre-existing condition lockouts that make them a no-go for our family. https://t.co/WRWKcR61Jv
It looks like these days most indie founders only care about making money as fast as they can. The majority don’t have the patience to stick with the same product for a whole year.
But that’s exactly what building a brand takes.
And that’s what unlocks the two channels you can’t buy that drive most sales for successful products. Word of mouth and being the answer when someone searches for a tool like yours.
I was one of the stupidest kids in high school.
I failed most of my GCSEs.
I got kicked out at 16.
I failed technical college.
I got a job as a baggage handler and it terrified me.
I worked really hard to get back into college and then uni.
I started building software products.
I've spent the last decade running my own software companies.
I made a over a million from it while travelling the world.
It doesn't matter how you start the race.
I love waking up to the smell of pre-sales in the morning ☺️
This is for my new Zero to Paid course on how to get your first customers.
I'm talking to everyone who pre-ordered over email and learning about their problems and how I can solve them with content. https://t.co/0rsxe4DjdG
My greatest skill is keeping the ship afloat in choppy waters with no fuel left.
11 years ago, a follow-on VC round for my company collapsed at the last minute.
We had real costs:
> employees
> contractors
> office space
> dedicated servers
and our runway completely disappeared overnight.
My co-founder and I suddenly had to figure out how to keep the company afloat without money.
We made a lot of uncomfortable decisions, cut costs and searched for new revenue streams, whatever it took.
Somehow we navigated through it but it took a long time and was deeply unpleasant.
It's easy to feel like a skilled founder when everything is working in your favour, but this is relatively easy.
The real skill is being able to navigate through the storm when everything is falling apart.
Building great software is still insanely difficult.
AI has made writing code faster, but code is only one layer of the problem.
Companies have had enormous engineering teams and virtually unlimited resources for decades and still shipped confusing, bloated products.
The hard problem is understanding what users actually need, deciding what not to build, and turning messy real-world problems into simple software.
AI cannot do any of that for us.
If you asked me to build a new product tomorrow, my first step wouldn't be opening Claude.
I'd find 10 potential users and get into a room with them or onto a call.
I'd want to understand how they solve the problem today, what frustrates them, what they've already tried, what they care about and what they don't.
Only then would I start deciding what to build, with those 10 users in the loop from day one.
AI can dramatically accelerate implementation. But if you don't understand your users or their problems, you're just building the wrong things faster.
It's easy to call someone a hater or a negative influence.
It's much harder to have empathy and help that person suceed.
I have been doing this long enough to know that success is not a destination.
Entrepreneurship is a never-ending rollercoaster.
I've had extreme highs and lows, some of those lows came after a decade of doing it.
I totally get why people get frustrated, jealous etc.
I'd rather be the person helping those who are struggling, not the person criticising them.
My DMs always open for a chat.
could never get around to HauntedFarms․com project, so I'm auctioning domain at Dropcatch․com; starting bid $15. Seems like a fun directory project easily built with ai. Offer free & premium listings. Goto Dropcatch and search 'HauntedFarms' (twitter doesnt like DC links fsr)
引用@searchboundacquired HauntedFarms•com at auction yesterday for $529 🎃
what sort of dolt would try building a (seasonal) directory for haunted farms?
danggit, me. d'oh.
I initially glanced over this name, but search volume showed I should take another look. … ↗
Dan Rowden@drindie105.4K粉 · 1条准备参加Indiecon，本期其余信号较少。
I'll be at Indiecon in Hamburg next weekend. An event I've been trying to get to for years. Super excited.
Can't wait to see lots of magazine people that I've only ever interacted with via email, video call or DMs for the past +10 years.
[Photo @Indiemags] https://t.co/EtsEVUAw55
▸ 折叠1条(转推/噪音)
噪音2026-08-26 Exciting!
Simon Høiberg@SimonHoibergindie161.5K粉 · 0条本期主要转发，原创信号较少。
转推2026-08-25 RT @SimonHoiberg: There's this fundamental flaw in all LLMs.
Even the best models do this. And it's probably the reason you can't make AI p…
Daniel Vassallo@dvassalloindie203.7K粉 · 0条转发HEY命令行与TUI发布。
▸ 折叠1条(转推/噪音)
转推2026-08-25 RT @jasonfried: HEY what!? HEY yes! The HEY CLI and TUI is out!
Calling all terminal dwellers, automation aficionados, and agents. Get you…