Britt Smith

If you think research is expensive or it takes a long time, just wait til you see what it’s like building products without it.

@elonmusk Ok but Grok is also giving LOTS of inaccurate information about the timeline of events. What’s worse? 😔

These are Microsoft employees

What’s going on here, @satyanadella?

@xdNiBoR 🙋🏻‍♀️🙋🏻‍♀️🙋🏻‍♀️

If you have access to the new XChat, you can join our space group chat where we can discuss anything about space! If you want to join, just reply here! You do need to setup your pin, and if you don't have access yet, just bookmark this and reply later when you have it!

@naval @elonmusk What use is it being a billionaire if it doesn’t buy you the opportunity to stand by your principles? Americans are so unused to seeing consistent ideology in politics, especially on the topic of national debt, they’d sooner believe this is conspiracy rather than philosophy.

Elon’s stance is principled. Trump’s stance is practical. Tech needs Republicans for the present. Republicans need Tech for the future. Drop the tax cuts, cut some pork, get the bill through.

@rrhoover all the internet OGs know zaps have always slapped 😆 they’ve always been innovative, stable, useful, etc exciting to see this chapter of “zapier makes you happier”

Playing with Zapier Agents. It slaps.

I was a total AI optimist until I found a 1k page conversation from a former employee who used to their work email to vent and plot revenge on the biz chatgpt over a 6 month period. It was super concerning from my pov how enabling the dialogue was and now I have serious concerns about this kind of interpersonal support from AI. I also felt like my consumption of the content (as the ceo) was pretty traumatic and like I’m stil stuck in an episode of black mirror. Maybe a good checks and balance flag might be something like “how could you be the problem here?” every couple hundred pages at least. I’m 300 pages into 1,000 and haven’t seen an example of the AI asking for self reflection, just straight up validation and encouragement, which makes me think the algorithm is more aligned for dopamine and engagement, rather than true self help or support, which is most often not the thing people in tough spots want (but need) to hear. It is what makes actual therapy both hard and valuable at the same time. Low quality therapy can be more harmful than no therapy for some people, in certain situations.

@alexgraveley A couple of things to consider: 1. It is phenomenal in that it can help many people - providing acccess, real assistance, etc.. 2. The data gathering without protections (with people sharing intimate information - both here & elsewhere) is concerning.

I’m calling it now. ChatGPT’s push towards AI assisted self-therapy and empathetic personalization is the greatest technological breakthrough in my lifetime (barring medicine). By that I mean it will create the most good in the world.

@GaryCardone Especially considering we typically have already paid for the R&D on the frontend (a tax if done w/ public funds). So yes, subsidizing the R&D and then paying more than other countries on top of that is insanity.

US citizens should have a 'favored nations' clause in all its pharmaceuticals, allowing US cititzens to NEVER pay more than the same drug issued in another country. It is ridiculous that we pay 5x more for drugs than say India or Mexico? Favored Nations Pricing Policy on ALL Pharma for US Cititzens using US pharma products.

Cursor is amazing but I also just said the other day one thing I miss about being in an office is that if you got to the point where you were crying about your code not working, there would always be someone around to take pity on you and help. It just feels too cringe to type out “I’m crying right now, seriously help” to AI. 😂

as someone who mostly learned to write bad code in a real world environment with a senior engineer growling at me…and later tried to go back and take a couple “foundational” CS classes at both Penn and Stanford learning new things with Cursor, Devin, Claude feels 💯 better

@teslaownersSV @elonmusk Twinkle twinkle little little stars, help America find its way to Mars. 🚀✨🎶

“I'll talk to Elon Musk. Elon, get those rocket ships going because we wanna reach Mars before the end of my term.”

@dharmesh Swagbot.io

LEAKED NEWS: I acquired the domain name os .ai for $150,000 (just saw the transaction reported on some online sources). It's the only two character domain name I own and one of the handful of .ai domains I've bought over the years. I have some ideas on what kind of startup/product would make sense for this, but I have no current plans to build anything (staying super busy with Agent.ai). The concept of an operating system for A.I. is compelling to me. I have a new agent on Agent.ai that I'm using to estimate the value of domain names. I think it overestimated the value on this one ($700k), but it got the key points/rationale pretty right. Also, the estimated value wasn't that far from the early asking price. If you have an ultra premium domain you want to get a valuation for, drop it in the comments and I'll run the new agent against it. (The new agent is slow, uses a lot of compute, and hasn't been fully tested yet, so I'm not ready to launch it to the public quite yet).

@dharmesh Lumineer.io

@elonmusk @DOGE phase 2, when the E starts standing for Effectiveness and we get all new success metrics for government that aren’t changed or obfuscated on political whim. 🇺🇸

A more accurate measure of GDP would exclude government spending. Otherwise, you can scale GDP artificially high by spending money on things that don’t make people’s lives better. For example, you could shift everyone who is building cars to working at the DMV. That would result in no cars and a much worse standard of living, but GDP would appear to be the same!

@paulg Agree, X has felt liminal for a bit now. But that also means more change is still coming…

Twitter seems a bit impoverished now that so many leftish users have fled. They may have been wronger on average about politics, but they were more interesting on average about other topics.

@jaltma Definitely see this happening with @notablyai and especially surprised (and excited!) to see how quickly some professors have been to get AI into the classrooms. Love to see it.

One of the interesting bits about this moment in AI is it seems like certain segments of the market have almost greenlit AI budgets without knowing what particular solution they want. Industries I historically would have thought of as laggards like healthcare, edu, defense, etc seem hungry to adopt AI and are incredibly receptive to AI products. This creates a little more risk of easy come easy go, but it's still noteworthy and largely positive for startups.

@jasonlk Offering great support is a reliable strategy to gather insights needed to move your software from mediocre to good to great.

Mediocre software usually also has mediocre support (at best). They seem to go hand-in-hand. But sometimes, you can be almost tricked into thinking mediocre software is good software if the support is amazing.

Our new integration with @MiroHQ is launching next week. We're getting sentimental on the blog about the role Miro has played in our research careers and how it helped to introduce design thinking to the world. More to come, but til then, a sneak peek. 👀✨ #AI #uxdesign #ux ##AI . .

So sweet to include all the babies born through the years and to see how Buffer supports families. Our team has a lot of long working history and relationships and it’s such a super power so I loved to see the data on retention too. Happy birthday, Buffer and congrats to you on this epic run as captain. Can’t wait to see what the next 13 years looks like. 🎂

13 years ago today, I launched Buffer. It’s hard to believe my little side project turned startup turned business is now a teenager. In recent years, I’ve reflected a lot on how to build Buffer to exist and thrive long-term. With that in mind, I thought it could be fun to share 13 fun facts which demonstrate our longevity mindset and the impact it has on both the product and the team. Here we go: 1. Over Buffer’s lifetime, our total revenue has been $159.16M. 2. We’ve raised only $4M, so we’ve made close to 40X our funding so far in revenue. 3. For 8 of our 13 years, we’ve been profitable. 4. 49% of the team have been at Buffer for over 5 years; that’s 39 of 79 teammates! 31 people (39%) have been at Buffer for over 7 years, and 5 people (6%) have been on the team for more than a decade. 5. 34 teammates have taken a 6+ week sabbatical since the program started in 2019. 6. 49 babies have been born to teammates in Buffer’s lifetime. 7. 4,955 active users started using Buffer over a decade ago, and 2,321 of those are paying customers. 8. We initially supported just 1 social network, now we have 11 channels you can connect with Buffer. 9. Over the last 6 years, we’ve spent $4.5M buying back the majority of our VC investors to build a long-term company with no exit plan. 10. Teammates have collectively received $1,395,923 through annual profit sharing distributions we’ve issued on 5 of our 13 years. 11. We’ve donated $269,185 to 28 different charitable organizations since our first contribution in 2017. 12. For 11 years, we’ve been a fully remote and distributed company. Today, we’re spread across 51 cities in 19 countries. 13. Finally, I’ve built and led Buffer working from 25 different countries over the last 13 years. Here’s to many more years 🥂

@NFT_GOD 🤩 Real-time data analysis is going to be so helpful for current events. I can’t wait to try it! Can we use an API to include with other sources for analysis in something like @notablyai?

Our team @notablyai is helping researchers across b2b, b2c, & healthcare… breaking into fortune 500s. Some customers include gen mills, news corp, Google, Harvard. Currently onboarding two traditional financial institutions. You have to be willing to go the distance with infosec being on the ground floor of new tech, but so far NRR shows it’s worth the effort.

The AI industry is in a curious state rn. Billions are being spent on capex, credits and tokens…yet few new incremental customer revenues are being generated - at least that I see. Two potential explanations: 1) AI is an efficiency play so companies will deploy agents and automation to reduce existing infrastructure/costs, keeping most end use cases the same. So AI generates OpEx savings “under the hood”. 2) VCs are feeding startups with billions that consume AI compute looking for new use cases. The AI compute complex booms and books real revenues but if the startups don’t find product-market fit soon then these revenues will shrivel because the startups won’t get more funding and will go bankrupt. What are some startups making money using AI to enable a new, valuable product to end customers/users?

@GregKamradt @SteveMoraco @OpenAI This kind of coding and the data visualization-ish, maybe even the testing itself (at least for gpt4) could be done with @notablyai 🙇🏻‍♀️

@SteveMoraco @OpenAI Automated, though per Anthropic's request they asked me not to use LangChain and instead go straight to them. So each model has mini optimizations the viz is very manual

This Claude 2.1 vs. GPT-4-Turbo chart absolutely blows my mind. Insane levels of fidelity from @OpenAI. Credit: @GregKamradt

@solysolz @emilychangtv @sama @adamdangelo @eshear For companies who operate within strict compliance environments… at what point does a key sub processor experiencing a coup trigger an incident response plan? Bc with some interpretations of security policies you could literally be required to spend time on this as a business.

@emilychangtv @sama @adamdangelo @eshear I want to sue the board for wasting my time

Update: - @Sama and the OpenAI board are now in talks for his possible return, specifically he's speaking with @adamdangelo - one scenario, transition board where Sam would serve as director - meantime @eshear has told folks he will leave if the board can't provide evidence of wrongdoing with @ashleevance @EdLudlow bloomberg.com/news/articles/…