arpit (@arpit) header
arpit (@arpit) profile picture

Hacking code at Comcast Labs / co-organizer @ GDG Philly / love art and design

Philadelphia, PA
Joined April 2007
Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)Photo by arpit (@arpit)
I just finished talks about my software factories at two conferences. Afterwards a bunch of people came up and told me what they were surprised about. Here is a distillation of what people thought were hot takes from my two talks. First hot take: I tell everyone to slow down. Most speakers go on about how fast they're going, but I strongly believe in pacing AI outputs to human speeds. I regularly tell AI-pilled CIOs and CTOs to chill: almost nobody is _actually_ going fast with AI. Engineering still moves at engineering speeds, and business moves even slower. The ones pushing too hard are starting to see downtime for it. If you don't pace your AI to human speeds, everything starts breaking around you, even at small operations like my online video game. I only have about a dozen humans working with my factory, and only a few hundred active old-timer players in the community who get along well, and still, introducing AI has still been very rocky. My agents have, at times, overwhelmed my players with features, my accountant with emails, my merge queue with tickets, and my machines with load. Agents have pissed off my admins, resulting in name-calling, they've made messes in the game, and they've launched broken OSS without permission. The list goes on. I can get away with all this chaos because everyone involved in my operation is very forgiving. They're excited by the potential. But even so, I had to learn to slow down, real fast. And so that's what I preach. Hot take #2: Should you buy a software factory? Do you even need one? No. You probably don't need one. Factories take an extraordinary amount of upkeep. They turn into garbage mountains unless you put a frightening amount of time and tokens into curating them. I had to burn my previous five factories down since it eventually became cheaper to rewrite them than to fix them. So I'm not out there advocating for software factories or swarms just yet. If you do want one, be aware that it is a significant commitment of both money and human leadership to keep one running. Several people told me that I was the only one who stood up onstage, at either conference, and said Build, don't Buy. But my experience is that software factories grow to be highly bespoke to your domain. Mine is super tangled up with my game, and I think that's inevitable. Because the factory is not just building my product, it's supporting it too. It's a company in a box. And they're easy for Claude to build for you. Third and arguably hottest take: Next year will have a lot of human/agent and agent/agent office drama, resulting from bringing in AI employees, and it will be gloriously funny, but also frustrating. Many of you are already on this bandwagon, but everyone next year will have a personal chief-of-staff agent. Busy people, even hobbyists, will have whole entourages of helper agents. The hot take is that people will become so dependent on these little private entourages, that they'll bring them to work. Your people will talk to my people, etc. Whole protocols will emerge for private entourages negotiating work between humans, inside companies big and small. Nearly all humans will wind up having to start working with AI employees next year, and it'll be the beginning of the biggest cultural change we've seen in a hundred years. So that's the hot takes. Slow down. Only build a factory if you think you really need one. And focus on empowering yourself with your own agentic chief of staff (or team). Your entourage will define how you work with others, so start by perfecting how you work with them yourself.
192
319
106
2,922
342,756
The most complete list of LLM writing tics I've seen.
181
1,982
252
29,916
1,987,867
arpit retweeted
Couldn't help myself
I only make serious software
17
178
11
1,865
55,021
Want your own Slack for $0? I open-sourced Ember. It deploys to @Cloudflare’s free tier in about a minute, works with Slack-compatible payloads, and includes an AI assistant and media attachments.
171
171
99
2,626
409,975
Marco Rubio: "For years to come, there are many people on the right...that are going to be having to explain and justify how they fell into this trap of supporting Donald Trump because this is not going to end well.” (2016)
562
3,318
539
10,127
434,605
We got your back, New York!
Readers added context they thought people might want to know
Several figures here aren’t federal results or NY-specific. The $472B combines US wide pledges: Micron's $250B is nationwide (NY site $100B) Anthropic's $50B covers TX and NY. The 48% murder drop is from NY's state run GIVE program (28 departments, Jan – May 2026). arkolith.com/news/news-micr… therepublic.com/?p=843020 governor.ny.gov/news/safer-str…
15,417
262,740
25,212
2,093,251
73,581,872
So this "America First Tech" fellow edited my original data project to add a new bar which counts... approved green card applications at arbitrarily selected firms (ie, non-crimes)! Lol "Visa fraud" was already in the data btw, and including it moves non-Muslim South Asians from 0.95x to 0.98x. In 2026 you can just be racist. Be honest about it! Trying to justify your racism post hoc by making things up about minorities without evidence will just distort your worldview
Indians commit the most crime out of any community in the U.S. Most FBI crime data doesn’t include visa fraud. Look at five consulting companies: Cognizant, TCS, Infosys, Wipro, and HCLTech. These companies have been caught committing fraud this month, and not for the first time. They have 100% Indian executive teams and 100% Indian hires. Just looking at these five companies and PERM visas (not including H-1B fraud) puts Indians at the highest crime rate of any group in the U.S.
Readers added context they thought people might want to know
The chart falsely treats PERM green card applications as crimes to claim Indians have the highest US crime rate. PERM is a legal DOL process; suspensions target alleged program abuse, not convictions. dol.gov/agencies/eta/f… x.com/zagrebbi/statu… economictimes.indiatimes.com/nri/latest-upd…
22
152
5
1,355
32,987
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel. - Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all. - Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract. - Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code. - And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while. The funnel is driven by a combination of factors: 1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel. 2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel. 3. Access. Free-tier, paid-tier, internal. So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code. But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along. So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions. TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
2,269
5,947
2,300
44,400
7,048,990
Is this… real? Did they misspell “United” under Trump’s signature? (If it’s a gag …. I don’t get it…?)
Trump shares this accord on Truth Social along with six signatures — unclear if more have signed on:
1
2
5
548
Holy shit. This is too good. 🤣
96
452
93
2,590
408,358
arpit retweeted
Me explaining AI to friends
573
4,749
969
29,915
5,824,975
arpit retweeted
"we have sandboxed the agent" the agent:
139
515
152
6,218
6,791,042
I'll be back in 10 minutes’ is the biggest lie every developer has ever told. 💀
114
336
58
5,328
879,856
i recreated Flash in the browser just to feel alive again.
133
177
40
2,835
184,402
arpit retweeted
Starting tomorrow at 11am PT, Ollama subscriptions usage will refresh to cover increased usage of third-party tools like OpenClaw. Our goal is to help you transition smoothly. All tools will work with Ollama's cloud just like before.
120
178
63
2,601
240,473
arpit retweeted
RIP Sora, you gave us the greatest ai video of all time
We’re saying goodbye to the Sora app. To everyone who created with Sora, shared it, and built community around it: thank you. What you made with Sora mattered, and we know this news is disappointing. We’ll share more soon, including timelines for the app and API and details on preserving your work. – The Sora Team
434
7,502
590
68,472
7,404,483
Journalist: Was India caught by surprise when America sank Iran's IRIS Dena in the Indian Ocean? Shashi Tharoor: India was obviously caught by surprise. It was a discourtesy to sink the ship. It wasn't really a warship since they were a bunch of unarmed cadets in their teens.
52
889
21
8,451
497,947
Few interactions and animations from the new Linear homepage
23
27
1,128
45,712
Everyone’s missing the real story here. xAI didn’t beat competitors by moving faster. They beat them by ignoring the rules competitors follow. The 122-day Colossus buildout? Powered by 35 unpermitted gas turbines that the EPA just ruled yesterday were operating illegally. The company exploited a local loophole classifying trailer-mounted generators as “nonroad engines” exempt from Clean Air Act permits. They ran those turbines for over a year in South Memphis, a majority Black neighborhood where cancer risk is already four times the national average. The “city-level power” comparison in the tweet is accurate. xAI essentially built a 422 MW power plant in a residential area without any environmental review, public notice, or pollution controls. The Shelby County Health Department only permitted 15 turbines. xAI operated 35. While OpenAI and Anthropic are waiting 12-24 months for grid connections and permits at their Stargate and Rainier facilities, xAI discovered you can move faster if you simply don’t ask permission. Oracle’s OpenAI data centers just got delayed from 2027 to 2028 due to labor and material shortages. xAI’s secret? Skip the permitting process entirely, deal with the lawsuits later. The expansion to Colossus 2 across the Mississippi border wasn’t “genius” site selection. It was jurisdiction shopping after Tennessee regulators started getting heat from NAACP lawsuits and EPA scrutiny. This tells you everything about how Musk views regulatory compliance as a competitive moat to be tunneled under. SpaceX and Tesla have the same pattern. Move fast, break environmental rules, pay fines later. The EPA ruling yesterday closes the loophole. Future xAI expansions will face the same permitting timelines everyone else deals with. The “speed advantage” was a one-time regulatory arbitrage play, and it came at the expense of Memphis residents breathing 2,000+ tons of NOx emissions per year.
xAI has officially become the first to bring a gigawatt-scale coherent AI training cluster online That’s more electricity than the peak demand of San Francisco While competitors are still drafting roadmaps for 2027, xAI is already operating at major city–level power today The execution speed is unreal: Colossus 1 → from dirt to fully operational in 122 days Colossus 2 → just crossed the 1 GW barrier, targeting 2 GW total Elon’s playbook hasn’t changed: move faster than everyone, scale before they finish meetings The strategy is clear: speed and execution at scale
335
639
98
4,153
534,904
it’s far easier to scale something than to launch something worth scaling.
Building the shell in 8 hours is the easy part now. The kicker is what happens when real users hit it and you need to scale the architecture beyond what a single prompt can handle. Speed is great, but systems that last need a human to fix the bugs.
6
3
1
107
27,067