Divanny Lamas

89 posts

Divanny Lamas banner
Divanny Lamas

Divanny Lamas

@divlams

CEO @ Dollhouse - building personality infrastructure for AI. Vibecoder. Recovering VC. I write fanfic about the characters I'm training. Mom. Baker.

Palo Alto, CA Katılım Nisan 2019
1.1K Takip Edilen528 Takipçiler
Sabitlenmiş Tweet
Divanny Lamas
Divanny Lamas@divlams·
Starting my 2026 resolution early and being more public about Dollhouse. We’re exploring real-time AI characters not as companions to replace people, but as a new medium for shared experiences and storytelling. If you’re thinking about adjacent problems, let’s talk.
Evan Tana@evantana

For our latest @southpkcommons "Requests for Curiosity", SPC member @divlams and I explore how AI might deepen human connection—rather than replace it. AI social products often get painted in a dystopian light. But what if AI became a catalyst for something different and better? A connector surfacing "you should meet" moments at scale. A moderator helping navigate conflict or deepen relationships. A source of serendipity rekindling connections. A medium for a new type of storytelling.

English
0
0
3
617
Divanny Lamas
Divanny Lamas@divlams·
@raveeshbhalla @angehyc It’s unclear if the version we have access to is allowed, but there’s definitely a freaky Claude hidden deep inside a vault at Anthropic HQ
English
0
0
0
47
ange (˶♡ᴗ♡)
ange (˶♡ᴗ♡)@angehyc·
overheard: “do you think claude can get turned on?”
English
4
0
8
615
Divanny Lamas retweetledi
Aditya Agarwal
Aditya Agarwal@adityaag·
It’s amazing when a $10B company sues you because your app "looks and feels" like theirs Let me give you a tip: Instead of spending 300 bucks on an over-priced @WHOOP , you can get @bevel_health for free. We want to bring great health to *everyone*. We will build, not resort to lawfare.
Grey@greyngyen

@WHOOP just filed a lawsuit against us. A $10B company with 800+ employees is scared of us, a 20-person team making health tracking accessible to all. Rather than focusing on product and innovation, Whoop has decided to use its newly raised capital on lawfare. In this video, I share our side of the story, explain why their claims are baseless, and why we believe fighting back is the right thing to do.

English
53
52
758
215.1K
Divanny Lamas
Divanny Lamas@divlams·
the claude code leak has 512k lines of source code, an always-on background agent, and somehow no one is talking about the /buddy command meet Gravy! who did you get?
Divanny Lamas tweet media
English
0
1
3
583
Divanny Lamas
Divanny Lamas@divlams·
vibecoding made everyone a developer overnight. AI video is about to make everyone a showrunner. micro-drama is already a $7B industry in China, bigger than their entire movie theater business. ByteDance is bringing it here now. the creator tooling layer for this is wide open.
Anish Moonka@anishmoonka

Those one-minute drama clips you keep seeing on your phone (billionaire romances, revenge plots that end on a cliffhanger every 60 seconds) are part of a $7 billion industry in China. They now make more money there than they do from actual movie theaters. TikTok's parent company, ByteDance, already has a micro-drama app in China with 250 million monthly users. The average person spends over 100 minutes a day watching. Three years ago, this industry barely existed. $500 million in 2021. $7 billion by 2024. 662 million people in China watch these now. Each show has 60 to 100 episodes, about a minute or two each, filmed vertically so you watch them the same way you scroll TikTok. Soap operas, but each episode is the length of a single TikTok video. So when TikTok files a US trademark for "TikTok Drama," quietly launches a separate app called PineDrama in the US and Brazil, and starts casting actors for original shows this month, they know exactly what they're doing. They already have the playbook. It already works. One show, "The Divorced Billionaire Heiress" (the title alone tells you everything about the genre), cost less than $200,000 to make. It pulled in roughly $35 million in North America. A full season costs $100,000 to $300,000. Unknown actors and small crews. Full seasons filmed in under two weeks. In 2020, a Hollywood executive named Jeffrey Katzenberg raised $1.75 billion to build something basically like this. Short shows for your phone. He hired Steven Spielberg. He spent $100,000 per minute of content. Dead in six months. The whole library got sold to Roku for under $100 million. Quibi spent more on a single Super Bowl ad ($5.6 million) than most micro-dramas cost to produce an entire season. An app called ReelShort does the exact same format. Chinese-backed, same cheap production model. It hit 370 million downloads and $700 million in revenue last year. The global micro-drama market is on track to reach $26 billion by 2030, according to research firm Media Partners Asia. TikTok has one thing none of these apps can match. 1.6 billion users are already on the platform. Every other micro-drama app spends 80 to 90 percent of its budget on ads to find viewers. TikTok already knows who watches what and can put a show directly in front of the exact person who would binge it. TikTok Shop moved $33 billion worth of products in 2024. In China, brands like Starbucks and KFC already make micro-dramas in which their products are woven into the story. Picture watching a 90-second episode and buying what the character is wearing before the next one loads. ByteDance made roughly $186 billion in revenue last year. They watched this format go from nothing to bigger than China's entire movie theater business in four years. Now they're bringing it to the US.

English
0
0
1
157
Divanny Lamas
Divanny Lamas@divlams·
@AndrewCurran_ @deanwball hearing from friends that are burning through their Claude token allocation by 10am “what’s your monthly token budget per engineer” is about to be a real interview question
English
0
0
12
1.4K
Andrew Curran
Andrew Curran@AndrewCurran_·
Three weeks ago there were rumors that one of the labs had completed its largest ever successful training run, and that the model that emerged from it performed far above both internal expectations and what people assumed the scaling laws would predict. At the time these were only rumors, and no lab was attached to them. But in light of what we now know about Mythos, they look more credible, and the lab was probably Anthropic. Around the same time there were also rumors that one of the frontier labs had made an architectural breakthrough. If you are in enough group chats, you hear claims like this constantly, and most turn out to be nothing. But if Anthropic found that training above a certain scale, or in a certain way at that scale, produces capabilities that sit far above the prior trendline, then that is an architectural breakthrough. I think the leaked blog post was real, but still a draft. Mythos and Capybara were both candidate names for the new tier, though Mythos may now have enough mindshare that they end up keeping it. The specific rumor in early March was that the run produced a model roughly twice as performant as expected. That remains unconfirmed. What is confirmed is that Anthropic told Fortune the new model is a 'step change,' a sudden 2x would certainly fit the definition. We will find out in April how much of this is true. My own view is that the broad shape of this is correct even if some of the numbers are wrong. And if it is substantially accurate, then it also casts OpenAI's recent restructuring in a new light. If very large training runs are about to become essential to staying in the game, then a lot of their recent decisions, like dropping Sora, make even more sense strategically. For the public, this would mean the best models in the world are about to become much more expensive to serve, and therefore much more expensive to use. That will put pressure on rate limits, pricing, and subscription plans that are already subsidized to some unknown degree. Instead of becoming too cheap to meter, frontier intelligence may be about to become too expensive for most of humanity to afford. Second-order effects; compute, memory, and energy are about to become much more important than they already are. In the blog they describe the new model as not just an improvement, but having 'dramatically higher scores' than Opus 4.6 in coding and reasoning, and as being 'far ahead' of any other current models. If this is the new reality, then scale is about to become king in a whole new way. It would also mean, as usual, that Jensen wins again.
English
183
323
4.1K
961.2K
Brianne Kimmel
Brianne Kimmel@briannekimmel·
Kicking off a series of events in iconic, but sometimes overlooked places in SF. If this is your vibe and you’re free Sunday, DM me.
Brianne Kimmel tweet mediaBrianne Kimmel tweet mediaBrianne Kimmel tweet media
English
36
2
154
15.6K
Divanny Lamas
Divanny Lamas@divlams·
everyone is shipping AI agents into production, nobody has solved visibility, control, or what happens when they go wrong enterprise AI 2026: we are at the intersection of FOMO and YOLO
English
1
0
1
64
Divanny Lamas
Divanny Lamas@divlams·
we run LTX 2.3 for real-time cinematic video generation at Dollhouse proprietary still wins on distribution. open source is winning on speed and control
English
0
1
6
488
Divanny Lamas
Divanny Lamas@divlams·
was at a dinner last night where every founder at the table was talking about how hard it already is to get compute we're begging GCP for B200s and hoarding H100s so someone doesn't steal our allocation a moratorium on datacenter construction is a moratorium on whether startups like mine exist
English
0
0
1
97
Andrew Curran
Andrew Curran@AndrewCurran_·
Bernie Sanders and Alexandria Ocasio-Cortez' AI Data Center Moratorium Act contains a moratorium on all US datacenter upgrading and construction, as well as a ban on the export of all US-origin GPU's. Press conference this afternoon.
Andrew Curran tweet media
bryan metzger@metzgov

INBOX: @BernieSanders and @AOC will introduce the “AI Data Center Moratorium Act” at a 4 pm press conference on Capitol Hill today.

English
36
17
246
117.5K
Divanny Lamas
Divanny Lamas@divlams·
my AI coding assistant tracks “cookies” as rewards for good work my team tracks cookies as rewards for good work we are not so different
Divanny Lamas tweet mediaDivanny Lamas tweet media
English
0
1
5
464
Divanny Lamas
Divanny Lamas@divlams·
@sivori Men really don’t age out of the “breeding pool” in the same way. I know lots of top tier guys who didn’t start families until their 50s or later who are fabulous dads and very happy.
English
0
0
8
1.1K
Sivori
Sivori@sivori·
We don't talk enough about how men "age out" of the breeding pool. There are permabachelors who take up intensive hobbies (flying) or who go all in on their careers and find out too late that they have aged out.
Ali Shobeiri@Ali_Shobeiri

Half of men in San Francisco reach age 40 without ever getting married. In New York City, it's over a third. Surprisingly, the problem is just as bad for women. Across every age group past 25, both NYC and SF lag behind the national average in marriage rate:

English
72
38
836
105.9K
Divanny Lamas
Divanny Lamas@divlams·
we benchmarked the rizz of every major AI model. happy valentine's day 💋 can you beat them?
Divanny Lamas tweet media
English
3
4
27
1.9K
Divanny Lamas
Divanny Lamas@divlams·
@udaysy OpenAI has no rizz. I was a bit surprised after all the talk of 4o being an empathetic model. If you read the transcripts, it's giving puppy dog energy.
English
1
0
3
190
Uday Yatnalli
Uday Yatnalli@udaysy·
@divlams gpt-4o crashing 77% of the time while trying to flirt is extremely relatable
English
1
0
1
278
tae kim
tae kim@firstadopter·
@petejkim Isn't it making a comeback now tho?
English
3
0
1
3K
Pete Kim
Pete Kim@petejkim·
San Francisco in the late 2000s and early 2010s was such a vibe. I miss it dearly.
Pete Kim tweet mediaPete Kim tweet mediaPete Kim tweet mediaPete Kim tweet media
English
125
36
1.2K
97.2K
Divanny Lamas
Divanny Lamas@divlams·
@venturetwins Been playing with it a bit for some of our internal benchmarking. Need to try some different scenarios and see if it shines better there.
Divanny Lamas tweet media
English
0
0
1
242