Daniela Amodei reveals the woman who trains Claude's character has a PhD in philosophy, not AI or ML
"It's unusual, right? You would think, oh, the person that's training the character of these models must be a computer scientist or somebody who's studied ML or AI for many decades."
"Amanda's technical and she technically trains the model, but I think it is very felt in Claude's character when you communicate with it that there's this kind of experience, something that's very wise, kind of the best version of the better angel of your nature trying to answer your questions."
"But if you kind of poke Claude in these difficult areas where maybe different philosophical traditions might conflict, I think that's where you especially feel Amanda because you'll sort of see like, 'oh well, you might want to consider things from this point of view but you could also consider it from this point of view.'"
"I think it's important because as the world begins to rely on these models more and more, you don't want to be locked into one philosophical tradition because ethics and morality are quite complicated, and different people and different cultures might have different answers about how they would like you to behave.”
A lot of the hardest technology problems don’t look like technology from the outside. Getting the right part to the right machine, on the right factory floor, for the right order, across nearly 7,000 suppliers, is one of them. @zetwerk made that work for customers in 29 countries, and then went on to build its own factories and take on harder products.
It’s been a privilege for Lightspeed to be part of their journey from the early days. Congratulations Amrit Acharya (@acharyaamrit), Srinath Ramakkrushnan (@srinathramki) and the whole Zetwerk team on the listing!
All this VC-Founder stuff is just a rinse repeat of stuff we see year after year in some form, a constant cycle and a lot of noise. Only one thing is forever.
You sleep in the bed you make. You make your choices for who you want to work with, the decisions you make and you learn to own it. Fail or succeed ... you own those choices and the losses and the wins they come with.
Everything else is meh .. get back to work already!
Today, with over 100 industry partners, we introduced the NVIDIA Open Agent Safety Platform, bringing together OpenShell and Sentry.
Artificial intelligence is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come.
But its full promise can only be realized when people have confidence that AI is being built to be safe and deployed with wisdom and responsibility.
This is bigger than a single product. It's the beginning of an open ecosystem to build the trust layer for safe agent systems.
Together, we are building the foundation of the AI economy.
Trust and innovation are not in conflict. Safety is how trust is earned. We must build not only the most capable AI, but the most trusted AI, so that this extraordinary technology can realize its enormous promise for the world. https://t.co/ugYWQ1MyRi
access to hardware is the biggest blocker in robotics. so we're sending robots worth $2k each to developers and creators.
learn, build, test and show us your work. apply now!
https://t.co/xwxN0ndp3Z
Unit X: How the Pentagon and Silicon Valley are Transforming the Future of War,” by Raj Shah and @chrismkirchhoff, is essential reading on technology and national security. Out today: https://t.co/oa0i50JjYp
easier to despise if you villainize.
@vkhosla once went on a month long family vacation in the middle of a 10 figure M&A (which he closed). he often took redeyes from asia just to make dinner.
he is an imperfect man but has an airtight circle and is the world's best father.
CHARGE 5-25X MORE FOR YOUR CONSUMER SUBSCRIPTION PRODUCT
Consumer CEOs: Charging $20 monthly / $200 annually for your product means that you need 1M subscribers to get to $200M ARR. Given the high churn with consumer SaaS products (4% per month not uncommon) you will need to constantly keep filling the top of funnel just to tread water.
Yes, ChatGPT and Claude have built large businesses with this low priced model, but they are the exception, not the rule. They've also each raised $5b+.
Consider instead, a plan where you charge ~$1000 per year, or even ~$5000 per year. Raising your prices by 5-25x does two things. First it, dramatically lowers the number of subscribers you need to get to real scale ($X00M ARR). As a side benefit, each subscriber is much more valuable, so you can invest in customer succes and retention. Second, it massively raises the bar on the value your product need to deliver.
Let's talk about the second thing. How do you raise the bar on value? You do it in four ways: (i) REFRAMING the role that your product plays in your customer's life (and finding a niche customer segment that cares about this) (ii) DELIVERING said value to this niche customer base, and (iii) COMPARING it to an amount the customer already pays for a comparable service.
Let's take an example. Calendaring systems like Calendly are either free for users or charge $10-15 per seat per month. However, what if such a system could reframe their product as an AI executive assistant, and subsequently build out this functionality? They can then compare the new price of $250 per month to the cheapest outsourced EA service (Eg: Athena) which charges $3000 per month. Obviously, the product needs to deliver value equivalent to an EA, but with AI, this is much more viable today than it was a few years ago.
One may argue that only a small % of people will pay $250 per month for an EA. But that's OK. Due to our higher per-customer revenue, our new system can afford to spend a much higher CAC to target and find 100,000 people (solopreneurs, SMB owners, coaches, etc) who will pay $3000 per year for this service. And also invest in a team to keep them happy and satisfied and not churn. 100,000 * $3000 = $300M per year! The economics become similar to a SMB-style business vs a consumer business.
So, to summarize:
1. Reframe the role of your service in your customer's life, ideally comparing it to an existing in-real-life service
2. Articulate a niche customer segment who cares deeply about this IRL service
3. Build this new, more expansive product that delivers this service
4. Price it much higher than earlier, comparing the value to what your target customer would pay for this service in an alternative world
5. Profit
Building a profit pool from a higher priced product will earn you the right to then build a new lower-priced product and go broader. This is the playbook that Tesla followed.
tldr with rare exceptions, I think most consumer products should start with a high priced version, dominate a niche customer base, and then over time, go broader with lower priced version. Most of them are doing the opposite, making survival hard.
absolutely thrilled to share our $12m seed round to truly empower everyone to create games and interactive experiences with AI.
a heartfelt thank you to our incredible investors AME Cloud Ventures, NVIDIA Ventures, Venture Reality Fund, Chaac Ventures, as well as the visionary leaders @ericschmidt , @NiccoloDeMasi , John Riccitiello, @DavidBaszucki , Jerry Yang, @mabb0tt , @silviocinguetta , and more supporting us in our journey.
https://t.co/DXyYVffCFu via @VentureBeat
A perspective on turning thirty from @sama : The days are long but the decades are short. Make it count, don't forget family, whether or not money can buy happiness, it can buy freedom, be around smart, interesting, ambitious people, learn voraciously. Also, lack of money is very stressful. Be a doer, not a talker. Exercise. Eat well. Sleep.
https://t.co/xQbLhYH5xo
This week we had 26 demo calls following our recent @ycombinator launch of https://t.co/wr42XG9ubt. We learned a ton of things by treating this demo calls as a customer discovery calls instead. 1/n
Hiring post for Chief of Staff:
@FoundersCareers is working with a leading hostel chain to hire an operations-focused generalist. Series A funded.
Work with COO and CMO to manage, streamline, and scale processes across various teams and verticals.
Form link in thread.
2015: Failed to get an internship abroad
2016: Applied a second time, and also failed
2017: Tried harder again, but still failed
2018: Finally became an intern at @Google NYC
2019/20: Got 5 full-time offers. COVID hit and the offer was canceled
2020: Started a deep tech startup
2022: Left the company. Applied to @ycombinator. Failed interviews
2023: Applied to YC again, failed again
2023: Launched Fluently: AI English Speaking coach
2024: Accepted into YC, moved to SF, raised over $2.5M in funding
The point? If you keep pushing, you'll get what you want. Sounds obvious, but it works 💪🏻
I'm ecstatic to announce our $22M Series A, led by @GoodwaterCap , with participation from @FirstMarkCap , @companyonvc and @epic_ventures.
This round will accelerate our mission to bring the world together through social experiences, empowering anyone to find targeted social experiences or start and scale an events community of their own.
Human connection is lost. POSH is the beacon guiding us back. It’s time to beat the loneliness epidemic, together.
Big, full circle news - I’ve rejoined @khoslaventures as a Venture Partner. I’m thrilled to work with @vkhosla@SamirKaul1@rabois and the entire team again, across enterprise AI, healthcare, infra and beyond. Excited to partner with the next generation of founders!
Last month I joined xAI as an AI Tutor Project Lead! This is truly the best in many ways!
We're currently hiring for roles across the full stack--check that out: https://t.co/k7w7rhsnIB
My team is notably looking for project leads, managers, and AI tutors (generalist, japanese-speaking, and coders) to support our human data program.
xAI is great in many ways, some of which are probably not super legible from the outside at the moment, so please reach out to me if you have any questions. The team is highly capable, we're up to great things, are likely to be 1 of the top companies that make it to AGI, have a good orientation towards safety, and we recently raised 6 billion dollars.