Iap .org / 1-Prct .com
Joined May 2023
- Tweets6.8K
- Following1.4K
- Followers80
- Likes17.9K
dv’s 2nd Account retweeted
Tuesday 10/27 (cont'd)
‣ An Invitation to the Seldon Speakeasy: A Séance luma.com/seldon-mwm6?utm_sou…
@higgsolivia
‣ Catie Cuan // Physical AI: Giving AI a Robot Body luma.com/hsmpoyxa
‣ Making First Contact: Space & Deep Tech Mixer luma.com/gb3rrrp9
@malayeriii
‣ How Partners Build on Lovable w/ Dan O'Leary luma.com/ff7g7tys
@franzschrepf @spc
‣ SF Systems: The Future of Data luma.com/osts92ax
@ShadajL @jakalinov @jeffreyhuber @cwensel @entersudonym
‣ Bay Area Biosecurity Lightning Talks 03 luma.com/s1ltny2j
@tessafyi
‣ Who Controls the Future of AI Coding? luma.com/dtzhn1ux
@openrouter
dv’s 2nd Account retweeted
Tuesday 10/27
‣ Lightyear Demo Day partiful.com/e/kRp3w7htyEthy…
@madhu___s
‣ Ping Pong Night // Inference × Depot luma.com/cygp5fj9
@ivictorialiao @samhogan @lizbethmetzler @depotdev
‣ Women in AI | Factory x Sapienne x Sequoia luma.com/factoryai-sdkw
@FactoryAI @theashleyjepson @francesca_lab @eliza_bara @maddiehfaulkner
‣ Startmate Accelerator Demo Day | San Francisco luma.com/SFStartmateDemoDay
@startmate
‣ Running Production with Agents w/ Rootly AI, NVIDIA, Replit, & WorkOS luma.com/r8sy88tq
@rootlyhq
‣ Founders (don't) cry luma.com/2eqnwxoy
@matteoc @frontiertower
dv’s 2nd Account retweeted
We're launching Greptile for firmware.
AI code reviewers are great for catching bugs in software, but when it comes to firmware, they miss critical bugs in ISR and peripheral timings, and runtime memory.
We're launching Simantic Bot to fix that. We gave AI the ability to run firmware in simulation, and test scenarios like providing external outputs from sensors and wireless communication, all inside your Github PRs.
Check it out at simantic.com/
dv’s 2nd Account retweeted
Anthropic’s IPO is what markets “have been waiting for, says Byron Deeter from Bessemer Venture.
He tells Ed Ludlow why 2027, post-Anthropic IPO, will be the year public market investors really get to participate in the AI wave bit.ly/4esWCty
dv’s 2nd Account retweeted
Anthropic’s long-term compute/infra contracts: $430B
China’s entire annual military spending (2025): $336B
but Anthropic’s $430B commitment is NONCANCELABLE,
meaning if the AI bubble pops, they lose their customers, or an asteroid hits Earth, they must pay the bill.
dv’s 2nd Account retweeted
being a founder means forgetting what day of the week it is
it’s always a build day. with a view tho!
dv’s 2nd Account retweeted
Probably no one will read this, but I will still write this as an important moment of our company @tryjamesai 's history.
This video was planned to be launched and run as ad next week, but I scrapped that plan today. This is not a launch video. We will change our product.
This James, this vision and version of it, had been created and will be used by our own team. But due to the limited time and money I have left to operate this startup, we are first focusing on more niches industries that we can more effectively advertise and sell to.
I was really attached to this vision, spent hundred of hours the past week to build the technology, and then refining the launch video countless time. But at the end of the day, after doing more research, I have to act accordingly to the truth no matters how I may feel.
In hindsight, I can be smarter, do more research and not wasted all these time building the wrong thing. But then if I didn't build this wrong thing, I may have not stumbled on our next idea, which I believe and hope it's better.
The truth is today, I have doubted myself.
The truth is I doubt myself a lot. On the outside, I may convey a sense that I will inevitably succeed, but it's more so because I can't accept a reality in which I don't succeed.
I have been going at this startup goal for the past 8 years, countless nights of working alone.
I will not give up now. I will make my desire into a reality through sheer will power and force.
dv’s 2nd Account retweeted
Daniel Kokotajlo risked 80% of his family's net worth rather than sign OpenAI's exit paperwork.
@DKokotajlo
dv’s 2nd Account retweeted
Adam Neumann agreed a $1.5 billion settlement with SoftBank, then his wife said no.
@AdamNeumann
dv’s 2nd Account retweeted
You have to hyperstition the future.
Have such deep set beliefs that you will make it that you forcibly carve reality to make it happen.
It's the only way.
Thiel talking about not backing founders who talk in probabilities and preferring those that speak in an absolute future.
Had to learn this lesson the hard way back when I was building and raising for my old startup.
I’ve made this mistake way too many times. And had to do a lot of changing to fix this. It’s too rational to be probabilistic. You have to be delusional and irrational to undertake a world changing company. Too rational ends up playing it too safe. That’s not what venture is about.
dv’s 2nd Account retweeted
What in the black mirror has Jack Dorsey done?
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.
block.xyz/dessa
dv’s 2nd Account retweeted
I made a free library of 100,000 historical clips 📚
search by the shot!
movingimagearchive.com
dv’s 2nd Account retweeted
not a raw transcript but the notes for the whole thing are up here with timestamps, your agent can just read that
readtube.co/videos/a-differe…
dv’s 2nd Account retweeted
HALLOWEEN PARTY. THURSDAY. BE THERE.
🎃
BOO! You better clear your plans for this one.
On October 29th, I'm co-hosting a Halloween Masquerade in SF alongside our friends at Photogenik.
We are bringing together founders, creators, investors, and the people building consumers for a glamorous night at a secret three-story loft…
Everyone will be masked and dressed up in their finest spooky formal attire.
This event will have:
- A hosted bar and DJ (featuring a special guest)
- A Photogenik photo booth
- Masks provided, or bring your own
Thank you to our friends at @rhobusiness, @pally (YC S25), and @humanpostco for helping make the night happen.
The guest list is capped at 150 so we’re curating the room and keeping it 21+.
Register using the link in the comments below.
dv’s 2nd Account retweeted
Moonshot deliberately weakened Kimi K3.1 Pro’s hacking ability before the open release.
the uncut version solved 90.5% of in-house exploit tasks, while the released model solved 81%, still well above K3’s 33.3% and behind DeepSeek V4.1-Flash at 85.7% by Moonshot’s count.
Moonshot AI’s first update, the new Kimi K3.1 Pro
> Kimi K3.1 Pro scores 65.7% on Terminal-Bench 4.0, 0.7 points behind Claude Opus 5.5 and 7.5 ahead of GPT-6 Astra, at a lower average cost per task than either rival’s top runs.
> Moonshot saw exploit skill climb as post-training scaled, so it removed cyber-related data from mid-training and RL; on its internal xdev-bench-hard, pass@3 fell from 90.5% to 81.0%.
> The filtered model still solves 81.0% of those exploitation tasks against K3’s 33.3%, which Moonshot reads as general coding gains carrying into cyber work, while a third-party run of GPT-5.6 Sol solved 100%.
> On Moonshot’s internal hallucination stress test with tools, false claims fell from 23.4% for K3 to 8.3%, still above GPT-6 Astra’s 5.5% but well below Claude Opus 5.5’s 15.0%.
> The biggest jump is on Moonshot’s simulated business-operation test, where scores rose from 8.9 to 37, and K3.1 Pro also tops OfficeQA Pro v2 at 64.4 against Opus 5.5’s 62.2.
> Gains are uneven: K3.1 Pro scores 38.1 on Terminal-Bench Science against GPT-6 Astra’s 68.1 and Opus 5.5’s 63.3, while leading Agents’ Last Exam at 43.2 and Humanity’s Last Exam with tools at 70.0.
> K3.1 Pro and a Highspeed variant running 2-3x faster are live in Kimi’s apps and API, with full weights due by November 4 for independent reruns.
