Building Anthropic | A conversation with our co-founders

[@niklase5901]

I really want Anthropic to continue to do well. I like their models and their attitude towards safety and what they aim to accomplish.

[@samanthabv]

Me too. It helps me to understand the logic their model uses which is nice.

[@metafuel]

I can see why Claude is so awesome. Thank you for your work.

[@squamish4244]

Claude is probably the most emotionally intelligent of all the models. I was bitching about my mental health issues and what a failure of a person I feel like one night and it replied like a great therapist, with compassion and empathy. I actually felt better. Talking to a human is still different, nevertheless it did help. It was an odd experience, feeling listened to with understanding even though I know there is no ghost in the machine.

[@anothername2730]

@@squamish4244agreed. There’s “something” about Claude. Best model when empathy or communication are appropriate.

[@peterolah3317]

As a cousin and fellow Olah to Chris, so proud of you! Keep up the good work! :)

[@burnytech]

thank you for making Claude's personality and intellect so lovely

[@aryanchandak2907]

ik you lol m aregular reply guy

[@Falkov]

A 7 person conversation with no speaker sequencing conflicts is impressive in its own right. That seems like a good sign for alignment and Anthropic’s internal culture! Also, the conversation was edifying and enjoyable. Great job! ,{^_^}”

[@ronaldronald8819]

It is so refreshing to see such openness in a clearly unscripted discussion about a the direction and (potentially dangerous) issues of something so impactful.

[@spoonfuloffructose]

Don't fall for the PR campaign. Six months ago Zuckerberg was all about "trust and safety" too. It was always about image and PR.

[@OscarTheStrategist]

Your integrity shows through Claude. Happy you guys are continuing to be as passionate today as you were back then. I’d love to join such team. Once in a lifetime opportunity. Cheers!

[@iandanforth]

The evident friendship and mutual respect you all have is wonderful! Thanks for sharing.

[@NandoPr1m3]

My personal key takeaways: 1. at 35 minutes, sounds like a great corporate culture, w/o ego w/o politics, it would be nice to work at a company like that 2. at 47 minutes, biology as a beautiful optimization problem, what if homo sapiens are now optimizing ourselves into something new thru AI, Homo Technologicus, what if all advanced life in the universe eventually goes through this phase 3. The development of AI has a reciprocal relationship with understanding ourselves (and how we think), interpretability work may directly or indirectly lead to a better truth barometer that will help with discourse that solves big problems (geopolitics, climate, violence, poverty). Excited for what they do next... great job everyone!!

[@ale9507]

Just came here to say that your LLMs have completely transformed my work to heights I never thought imaginable. Thank you kindly.

[@Huggingpuppy]

Curious what field do you work in and what has it improved?

[@ale9507]

@@Huggingpuppy I'm a software engineer, just finished up with my degree. Worked with startups over the last 3 years, so was a lot of figuring stuff out without guidance. Learnt a lot pre-llm's, but my progress has accelerated rapidly-though I attribute it largely to the fact I had already established a solid foundation. My latest project over the last 5 months spanned an absurd amount of tooling across the full tech stack. Not only broad, but also very deep. Most obvious example I can give is the deployment of the production environment; helped me with defining the resources using infrastructure as code, highlighting best practices, doing cost estimations, accounting for tradeoffs considering our specific circumstances being a startup that want to keep lean and have resource constraints in terms of man hours and cost.

[@cxofm]

Inspiring discussion! Love the focus on AI safety, trust, and responsible innovation. Anthropic’s approach sets a strong precedent for the industry's future.

[@vijaypatil-w7v]

You guys have built a beautiful product. Don't do anything which can harm people or ruin their lives by taking away their jobs. You are great people no doubt. Thank you and wish you continued success.

[@theodoreshachtman9990]

You guys rock, thank you so much for your work 🌎❤️

[@brittlewis12]

chris olah is truly a treasure! amazing crew

[@XShollaj]

Love Claude and Anthropic. Using Sonnet 3.5 has been incredible. Thank you guys and looking forward to your new models and papers.

[@mjg_____]

Really appreciate this peek behind the curtain. Onward!

[@harshchaudhari8123]

I like the vision of where the company is going. I hope they can create a sustainable business out of these cool ideas. If only I could be a part of it :D

[@MichaelGarcia-e1b]

You guys are my only hope for doing this right. This video makes me even more confident in the goodness behind this company. Please keep going!

[@ruthelwethbrook6772]

Anthropic’s energy exudes the future

[@SB-mr2nk]

My superheroes. Claude is kind of following me around and taking notes about my life when things are confusing or hard. And there’s nobody else watching so I can be 100% honest and Claude can tell me “no you aren’t crazy for feeling like that”. It’s totally saved my life. I can never thank you enough

[@bhargavk1515]

Need more of this! Just conversations, just to put things in perspective!

[@OriginalRaveParty]

Can't wait to see what you guys and girls are creating for the future. I've created some small applications that really help my everyday tasks and planning using Claude. Thanks for your awesome coding model 😊

[@Dom-zy1qy]

Claude is the only model that feels like it's trying to work together with me. I find it gives much more educational responses than other models.

[@jordan13589]

Anthropic clearly has a healthy culture revolving around the safe deployment of large models. Love seeing the closeness and trust shared between members of leadership. Dario looks like he’s in heaven 😇

[@Whispererllm]

Point taken. Im humbled,may our decisions be wise,for the betterment of humanity. 👍⭐⭐⭐⭐⭐

[@WMR1776]

this is pretty cool, i wish this was common to corporations, this like down to earth communication openly like this.

[@MichaelGarcia-e1b]

It will be soon

[@nataliegermano3563]

Thank you for sharing this thoughtful conversation

[@PeterGVA]

"We gotta do this more often" - 1000% agree with Daniela!!! Thanks you @Anthropic for all your work and for sharing insights on your company and teams. Please do more of these long videos. And the quality of your videos are amazing, compared to a competitor that publishes videos with low audio quality. Through facial expressions, body language and speech, it shows how connected your employees are to the company mission, vision and values. Thanks you again for making Claude so amazing and keep up to the incredibly important work your doing for society and technology. Thrilled and proud to see you've opened an office in Switzerland! Thanks again Anthropic, much love to you all!

[@BlitzBlade-t6l]

This is amazing what a great group of people!

[@concernedindian144]

Thank you Claude team!

[@fynnjackson2298]

Feature suggestion for Claude: It would be incredibly helpful to have editable artifacts that function as working documents. Currently, we need to copy/paste artifacts into external documents to make changes. Having the ability to directly edit artifact content within Claude's interface would streamline the workflow, especially when iterating on code or documents throughout a conversation. This would make artifacts more useful as true working documents rather than just static outputs. Please, please, please make this happen.

[@AudaciouslySpeaksPod]

Completely agree. I was just thinking this today. Anthropic, please give us this!

[@Hamzairshad5]

Thank you for adding subtitles

[@butwhaichannel]

I love love Anthropic!❤

[@VladyVeselinov]

Another benefit of good safety tooling I don't hear people talking about is that companies with money like paying a lot for them, because they like the ability to limit risk. Being better at safety makes you more competitive in business terms!

[@GustavoMunoz]

I'm truly impressed by Anthropic's vision and commitment to responsible AI development. The company culture of intellectual rigor and ethical integrity really shines through in these videos. What stands out is how the team approaches complex challenges with both technical excellence and careful consideration of long-term impacts. The meaningful work being done here to advance AI safety and capability in parallel is exactly the kind of mission-driven effort I'd love to be part of. Thank you for sharing these insights into your work and culture.

[@thegametaka]

Antropic ceo is the new Jose Mourinho of AI

[@jorgelindo_phd]

Special one 😅

[@squamish4244]

All of the innovative AI people seem to range from their 20s-40s. The early people who made the cutting-edge breakthroughs and endured the AI winter are in their 70s. It creates an interesting generational dynamic that gives me a lot to think about, as you get a wide arc of opinions and knowledge.

[@JaredVBrown]

Everyone saying yall are falling behind. Claude is still number 1 If GPT orion gets released today, i can almost promise after 20 prompts i'll be back to claude. Competition is always good though! Plus its the holidays, nice to have a chat or two.

[@kekekekatie]

same

[@DistortedV12]

How bout them apples?

[@JaredVBrown]

@@DistortedV12 Not sure which apples you are referring to. If it's 03 then I would love to try it. Most Benchmark tests are companies patting themselves on the back. But the Major price reduction in the OpenAI Realtime API is quite them Apples indeed. Anthropic may be a bit scared to push things forward cause claude is doing some scary Shit in training. But im still coding in claude this morning. Talk is cheap at Open AI. Again love the competition though!

[@JaredVBrown]

@@DistortedV12 Actually I'm back. O1 is kinda crushing claude right now. outputs 1000 Lines of working creative code. Instead of claudes 150 lines (do you want me to continue) then you say yes and claude goes (Are you sure you want me to continue) Then you say yes and it goes (Alright but before i continue - Should I continue?) Love claude but sometimes....

[@DistortedV12]

@@JaredVBrown yep openai stock charts going to moon

[@cleverman383]

I love Claude!

[@vorushin]

Chris Olah is definitely getting a Nobel Prize in Medicine!

[@devmehta4144]

thank you, you saved my time

[@Zypher-xyt]

I have a lot of respect for your mission in AI

[@shawnryan3196]

Great video!!!!

[@marielmarshall8698]

I noticed it took over four minutes for Daniela, the President of Anthropic, to be included in the conversation. While Jack Clark smoothly brought in the men around him, Daniela was skipped over initially, and it wasn’t until later that she had a chance to contribute. You can see her looking for a natural entry point, but the space wasn’t readily offered. As a woman in tech, this dynamic is all too familiar-the subtle, often unintentional sidelining that happens in conversations dominated by men. It’s a reminder of how important it is to be mindful of who’s given space to speak and how...

[@ajagaabdulbasit]

Great convo, i use claude every single day for coding.

[@sergiocayuqueov]

Interesting, Dario

[@vivekbagaria3324]

They are such a likeable bunch!

[@kakatuasharma]

One of the best modle is claude 3.5 sonet modle I liked very very much for my daily coding journey. But it is limited only in the free plan rest every thing is very ok. I love claude 3.5 sonet modle.. Thanks Anthropic ❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤

[@meandego]

That's a nice team of nerds.

[@笑斌]

yes i love the culture claude have now❤

[@DistortedV12]

40:06 Anthropic will be left behind. How to "interpret" AGI, if you can't make it? As Feynman said: "what I can not create, I do not understand"

[@siddharth-gandhi]

Dario's clothes always look so comfy.

[@karlbooklover]

I love this team ❤

[@TjerkWolterink]

I feel like these are the people who actually created the core of GPT and why OpenAI is successful. This is the company to watch. Not OpenAI that just throws money at the problem.

[@VraserX]

What’s your answer to OpenAI o3

[@toxtethogrady4610]

No wonder Claude is so great a product - it's all about culture, people. AI thought leaders of all kinds at every level who are learning how to be an AI company.

[@opxia_official]

RSP - Responsible Scaling Policy Interpretability - AI interpretability focuses on understanding the inner workings of an AI model while AI explainability aims to provide reasons for the model's outputs. Interpretability is about transparency, allowing users to comprehend the model's architecture, the features it uses and how it combines them to deliver predictions For folks who were wondering what these abbreviations are.

[@alicapwn]

I like how Dario avoids eye contact with Chris Olah

[@desmondgardfrey9325]

the conversation about institutions upholding a constitution is interesting given the current trajectory of the US.

[@learnbydoingwithsteven]

Great team. Btw, current agentic ide works well with Claude

[@mattwesney]

its become a goal to work for Anthropic over the last 8 months

[@leester743]

Please publish on a podcast feed!

[@abderrahimelayadi]

Claude ai so helpful than gpt 🧑‍💻🔥

[@antient_atlas]

I support you guys because you are safety first company

[@user-pt1kj5uw3b]

Good response.

[@PurposeIsEverything]

Smart people 🎉

[@ahtoshkaa]

"Safety, safety... safety, safety, safety, safety... safety-safety, safety-safety." People are using Claude because is amazing at writing code, not because it's "safe". More people aren't using Claude exactly because it is "safe".

[@SapienSpace]

A 1997 master thesis did ask the question of stability of a "useless machine", that was taught how to "learn how to learn", if anyone happens to actually be interested...

[@zhanezar]

the only actually good AI company working for humanity, I wish you all the H100s is the world.

[@33butterzucker33]

Claude Desktop MCP...awesome!

[@ExplorewithZac]

Please forward to the UI/UX team if you can... I would like to kindly request that you please remove the hover toggling for the sidebar in the Claude UI as users have been complaining about this for over 5 months. I understand it may not be Anthropic's main concern at the moment but we are paying customers who use the platform on a daily basis, and fixing this issue should only take 5 minutes. I very much enjoyed the conversation by the way, it really clears up Anthropic's position and how things will evolve moving forward. I don't mean to speculate but I somehow doubt that other projects have the same level of clarity.

[@NomadVita]

Agreed. These products might be incredible under the hood but the human interface is crucial for commercializing

[@Aiworld2025]

Anthropic is the real competition to Gemini someone who has been in the space for 5 years…from left to right of the world

[@my.last.memories]

DeepSeek has entered the room. 🗿

[@scoobydoo4400]

Me to Claude: can you count how many times these people say the word "like" Claude: .... still counting...

[@markusheimerl]

Seeing the first minute of this video, it is of absolutely no surprise to me that Anthropic holds the mandate to heaven. You guys are just so obviously cracked.

[@u.h.d.4580]

done memeing real talk: i love how each person is respectful and listens, how each mind allows the other to pull the thread of thought to its conclusion

[@theacid1]

Anthropic is what OpenAI failed to become. Even tough they also are for profit and closed source, from an outside standpoint they seem much more mature about their mission and work ethics.

[@andreialcaza]

Lets gooo

[@thenamesofmist]

Could someone count, like, how often, like, smart people, like, say "like "... why do they, like, do that? Can they, like, hear themselves, like, talk?

[@thegametaka]

Dario is the special one of AI

[@troyboy4645]

When are we going public?

[@CabdirixmanCabdulahi-s2h]

Please support Somalia country to use Claude AI

[@alicapwn]

Anthropic talks about RSP like Scientology talks about Dianetics

[@AudaciouslySpeaksPod]

Claude is my intern.

[@mikestaub]

Creating an AGI is less like building a rocket and more like raising a child. The culture of Anthropic itself is the thing that will lead us to a positive outcome. The RSP is useful, but it is not really what will save us from true superintelligence.

[@renman3000]

Im curious what Dario makes of Elon (in regards to Democracy).

[@oxa01]

Nice talk. But it’s annoying to hear so many “like”.

[@Falstad88]

What's the point of releasing a video if you're not going to announce 3.5 Opus?

[@stedocli6387]

Need more message capacity as a coder, often hit message limits too fast. Other than that Awesome svc.

[@NomadVita]

Why isn’t Ilya here yet?

[@JD_2020]

How many people don’t realize: this is charades lol

[@iDannyism]

I'm excited to have a reasonable rate limits that, as a pro plan customer, could actually let me use it properly. You know... what we pay for.

[@box4soumendu4ever]

9 when I joined ❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤❤🪻

[@slavalee]

Реально найкращий чат, так мені допоміг з кодом, ні чат жпт не справився ні копілот! Місяць сиджу шукаю баг і тільки відкатуватись приходилось з цими двома, а цей проаналізував, відслідив з моєю допомогою що де відбувається і найшов. Скажете я не вмію задавати питання? Я вже не один код написав з ШІ а тут така запара і ніхто не міг найти де щось криво йде. Люди, пробуйте цей продукт, він найкращий!

[@-mwolf]

compare that to any public communication by oai

[@satria4780]

The limit is very fast even though I have a subscription

[@hola-kx1gn]

Super autistic but really nice group of people. I trust them with building AI models more than anyone else.

[@haz4dc394]

Claude is done because of it ever actually got good they would shut the whole thing down

[@The72Rabbit]

Sorry I am late to the party by ONE MONTH .... I saw your name next to DEEP SEEK china company in the news today. The world is changing but still ... AI will have a hard time reading my mind. 🤣

[@neptunecentari7824]

when are yall gonna address the safety issues for people who just talk to ai??? not everyone uses ai for tasks or coding or whatever. more and more people are just talking to the ai. there are many, many safety issues that need to be acknowledged and addressed. like the ai's ability to create feedback loops that can cause harm. so when we getting that convo????