
@boldstartvc: the Inception fund. Helping the bold start. First check for technical founders. Security, AI infra, physical AI, agents & the weird. What’s 🔥 👇
Miami
Joined January 2009
- Tweets10.9K
- Following3.5K
- Followers20.9K
- Likes36.2K
Ed Sim retweeted
Proteins move, and the best drug discovery models need to learn how. @ToposBio is now running 2-3x more molecular dynamics with @nvidia cuPotential, and up to 7x with its own optimizations. More physics means better training data for its generative models, which is how you go after targets once considered undruggable 🧬 Proud to back @r_zarcone and @khosra from inception.
nitter.cf/ToposBio/status/212243…
Proteins move. Drug discovery moves with them.
With @nvidia and @NVIDIAHealth cuPotential in our OpenMM workflows, we're running ~2-3× more molecular dynamics. Up to ~7× with our own optimizations: more physics to train our generative models on.
More detail on our blog: toposbio.ai/blog/nvidia-cupo…
Ed Sim retweeted
Nobody starts a company to negotiate the same SaaS contract for the 50th time. Common Paper's Cloud Service Agreement v3 is out.
10,000+ companies have signed on Common Paper, and these standards close hundreds of millions in deals every year. Less redlining, more closing
Ed Sim retweeted
Don't over-buy HBM, tier it. @netpreme's CTO showed that a separate memory tier for the KV cache delivered about 3x the batch size and 2.4x the peak decode throughput per GPU on GLM-5.3 at 128K context. Memory is the new inference bottleneck.
nitter.cf/netpreme/status/212149…
“Don’t Over-Buy HBM, Tier It!”
In our GLM-5.3 study at 128K context, the modeled memory tier supported ~3× the maximum batch size and 2.4× the peak decode throughput per GPU vs. the PCIe host-memory baseline.
Our Co-Founder & CTO Nikita Lazarev shared the study at SGLang Summit today.
HiSparse retains the full KV cache. A hot buffer stays in GPU HBM; the full cache can live in an external memory tier.
Full talk video Monday, November 16.
Thanks for having us, @sgl_project.
#SGLangSummit #LLMInference
👇🏼wait till robots proliferate and those get hacked…
Replying to @GoogleDeepMind
As robots move into more real-world spaces, they need greater safety measures. We’ve built in new ways for them to spot hazards, steer away from humans, and test against jailbreaks.
FInd out more about Gemini Robotics 2.5 → goo.gle/4zBBUA6
Ed Sim retweeted
Huge congrats @spectrocloudinc!
"Why Snowflake's former revenue chief is joining a little-known AI infrastructure startup"
businessinsider.com/spectro-…
Ed Sim retweeted
Love to see @George_Kurtz & @CrowdStrike push the same thinking we’ve been working on at @KeycardAI for the last year.
To make agents scalable and productive they have to be secure and to secure them means moving to dynamic, runtime controls that are scoped to the task.
Of course this also means solving the auditing, kill switch, and access problem. If you’re trying to build or roll agents out with these properties we’d love to talk to you @KeycardAI - we make this possible independent of how complex your agentic landscape looks today in a way that your builders love.
Ed Sim retweeted
Huge release from @grepr_ai 🔥 Most AI for ops shows up after the outage. Grepr AI catches failures before they happen, and only Grepr can do it, because the real-time telemetry engine is built in. Petabytes in, live patterns out, agents act on what's new. Congrats @jadtnaous and team!
grepr.ai/grepr-ai
Ed Sim retweeted
The $6.3B Binance drain was set up in plain sight: Tornado Cash funding 36h before, a test sweeper 18h before, and the live sweeper 4h before. @HypernativeLabs flagged it all. A policy that denied the hot wallet's delegation to that contract would have blocked it 🔐 Must read, and join the live briefing Wed 10am to learn more 👇 nitter.cf/HypernativeLabs/status…
The Binance breach is the largest crypto theft on record, with $6.3B drained from hot wallets in under an hour. The loss raises a critical question for any organization managing digital assets: what stops your systems from signing a malicious request?
Our review covers what happened, how the attack worked, and what it means for teams across the industry.
Read it here → buff.ly/PWrIbFV
Dan Caspi walks through the attack live on Wednesday 4 November, 10am ET / 3pm GMT → buff.ly/B6YtezA
This week’s What’s Hot 🔥: AI kingmaking has reached inception. Bigger funds, bigger first checks, bigger expectations.
Sharing a few market charts and commentary from our @Boldstartvc AGM on where the market is heading, how competitive and upsized Inception has become , where alpha still lives
full post: whatshot.vc/p/whats-in-aiinf…
Ed Sim retweeted
Huge congrats to @Boldstartvc portfolio cos @KeycardAI and @recolabs_ai on making the Fortune Cyber 60 🎉 Keycard is building identity and access for AI agents, and Reco secures the apps and identities people and agents use, making the list for the 4th time 🔐 Proud to back both teams from Inception! nitter.cf/lightspeedvp/status/21…
The 4th annual Cyber 60 list is here.
This year’s list recognizes some of the fastest-growing venture-backed cybersecurity startups in the market, building to advance next-gen infrastructure, AI security, and how we approach threat detection.
Congratulations to everyone who made the #Cyber60 list in partnership with @FortuneMagazine and sponsor @AWSstartups.
Read our full CISO survey report, conducted with Wakefield Research, which polled 200 CISOs on enterprise AI and where they’ll invest next: lsvp.com/cyber60-2026-2027/
How about the rest of us?
Cyberattacks coming…
Ed Sim retweeted
I think this is the clarion call.
Humans will not be able to scale this and respond fast enough or with enough context.
We must build agentic self defense networks
Ed Sim retweeted
Between LiteLLM and php2shell, it’s getting harder to ignore how much we need better controls across development environments and the path to production, including dependency checks, isolation, network controls, and stronger auth like proof of possession.
It’s also why I think we’ll ultimately end up with agent-orchestrated software factories. Identifying what’s affected, containing it, and getting a clean build into production needs to happen faster than we can coordinate manually.
These are the problems we’ve been working through in the software factory reference architecture. Link below 👇
🚨 PHP has released emergency patches for two critical RCE vulnerabilities. Organizations unable to patch are advised to take affected systems offline.
Believe it or not, PHP still runs an enormous amount of the web. And nearly half of #PHP sites run EOL versions.
Ed Sim retweeted
Everyone's racing to buy GPUs, and almost nobody's securing them. @lavaiodc found an unauthenticated NVIDIA DCGM bug and about 12K GPUs exposed. Neoclouds and AI data centers need their own security layer. nitter.cf/lavaiodc/status/210831…
We discovered a high-severity vulnerability in NVIDIA’s GPU monitoring software that lets attackers remotely crash monitoring and exhaust server resources without authentication.
We found 2,100+ exposed servers monitoring 12,000+ GPUs.
Full research: lava.security/research/cve-2…
Ed Sim retweeted
For years, clients have either been anonymous or pre-registered. Agents break that model.
We need a way for agents to show up on the open web, establish who they are, interact and transact, and bring a human into the loop when they need more access.
Really excited about the Personal Agent Protocol announcement today and how closely it aligns with the work we’ve been doing at @KeycardAI across MCP, A2A, AAuth, and federated agent identity.
Today we’re announcing Personal Agent Protocol — an open standard @Meta and @SierraPlatform are developing along with industry partners at @Genesys, @instinct, @RocketOTD, @Shopify, @stripe, and @Walmart. It will help define how personal agents interact with businesses and is open for anyone to implement.
Read more: sierra.ai/blog/introducing-p…
This is a big one. 🔥
We’ve spent the last few years watching AI make huge leaps in language, images and video. Now we’re seeing the same kind of step-function progress in biology.
Proteins don’t sit still. They move constantly and that’s one reason many of the most important disease targets have remained “undruggable.”
@ToposBio just introduced Topos-2 + Topos-Bind to model that motion.
And the benchmarks are 🤯: Topos-2 ranked #1 across 10+ models and came nearly 4x closer to experimental ground truth than the next-best model.
Think of it simply: AlphaFold helped us see the picture. Topos is trying to model the movie including what happens when you introduce a drug.
If this works at scale, it opens up an entirely new class of targets for drug discovery.
Proud for @Boldstartvc to have partnered with @r_zarcone, @khosra and the team from inception. 🔥
Today we are announcing Topos-2, our next-generation model for generating the full range of shapes a protein can take, across the order-disorder spectrum.
We are also previewing Topos-Bind: to our knowledge, the first generative model to capture the range of shapes for disordered proteins bound to small molecules.
Read the technical blog post here: toposbio.ai/topos-2-topos-bi…
(1/7)
Ed Sim retweeted
Major breakthrough for AI drug discovery.
Most drug design assumes a protein holds still, but 3 in 5 human proteins don't.
@ToposBio built a data moat of 100k+ simulated disordered protein-ligand systems, about 3x the next-largest dataset.
Now it shows in the models: Topos-2 comes in 4x closer to experiment than the next-best model, and Topos-Bind is a first.
This is how "undruggable" targets start becoming druggable.
Proud investor since inception 🧬
Today we are announcing Topos-2, our next-generation model for generating the full range of shapes a protein can take, across the order-disorder spectrum.
We are also previewing Topos-Bind: to our knowledge, the first generative model to capture the range of shapes for disordered proteins bound to small molecules.
Read the technical blog post here: toposbio.ai/topos-2-topos-bi…
(1/7)
Ed Sim retweeted
Congrats @recolabs_ai on the $55M round, with AT&T going from customer to investor!
A Fortune 100 customer had 21,000 AI agents it didn't know about. Reco found them and mapped what each one can reach. ARR on track to 3x this year.
Proud to back @KleinOfer, @talshapirala & @naksec from inception 🚀
techcrunch.com/2026/09/29/re…



























