the inception fund. helping the bold start. first check for technical founders. Cybersecurity, AI infrastructure, physical AI, agents, the weird. since 2010
Miami
- Tweets 4,595
- Following 663
- Followers 8,845
- Likes 7,139
🤯 world models, wild money 💰
@AMD just bought @theworldlabs for $8.2B
@gen_intuition raised $220M at $6.2B
Field AI reportedly raising $700M at $10B
founded 2023–2025, first checks of $50–100M, now worth $6–10B
everyone's racing to turn model capability into useful action in the physical world… so so early but still so much more value to be unlocked
What's 🔥 #518 👇🏻
whatshot.vc/p/whats-in-aiinf…
2
3
12
1,735
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)
2
5
13
1,621
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)
2
3
801
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…
1
2
4
493
The best signal in enterprise security is a customer who writes a check.
AT&T went from @recolabs_ai customer to investor.
Agents are getting keys to every app, and Reco maps what each one can touch.
Huge congrats on the $50M round.
Proud to be with @KleinOfer, @talshapirala & @naksec from inception 🚀
Reco raises $50 million with AT&T backing to secure AI agents.
The company, which began working with AT&T as a customer, plans to expand sales, partnerships and customer support as large enterprises deploy more agents.
calcalistech.com/ctechnews/a…
1
1
2
535
boldstart ventures retweeted
Introducing Quotient Analytics.
Most marketing analytics starts after the work is done.
You plan the campaign in one tool. Create the email and content in a few more. Publish social somewhere else. Then, once it is all live, you try to piece together what happened across GA4, email reports, social dashboards, and your CRM.
That is backward.
Marketing needs a tighter feedback loop: plan the work, ship it, see what it drove, and make the next campaign better.
Quotient Analytics brings that loop into the same platform where the work happens. It connects email, social, and website activity to the campaigns that drove it, with automatic tagging and an agent you can ask for answers.
The goal is not another dashboard. It is a marketing system that learns from the work you put into it.
Go check it out: getquotient.ai/features/anal…
3
4
11
586
Meta’s Muse is the easy button consumers have been waiting for 🔥
Fun, intuitive, and actually doing things for you.
Now unleash tens of millions of them shopping, booking and calling 24/7 🤯
Who secures them? Whose systems survive the chaos?
Read this week's What’s 🔥 #517 : whatshot.vc/p/whats-in-aiinf…
2
3
8
3,107
This is what a real inference breakthrough looks like. 🔥
@Systalyze is now #1 on Artificial Analysis across every model configuration it entered.
Gemma: 461 tok/s vs 109 for #2. DeepSeek V4 Pro: 274 vs 191.
Systalyze is attacking inference as a full-stack systems problem — from model execution down through the hardware.
It just works at blazing speed across very different models!
@systalyze is now #1 in output speed on @ArtificialAnlys, on every model we entered.
→ Gemma 4 26B: ~10× faster than Google
→ DeepSeek V4 Pro: 4× faster than DeepSeek
→ Muse Glimmer: 2.5× faster than Fireworks
Try it via our OpenAI-compatible API:
console.systalyze.com
5
3
15
3,410
AGI isn’t the bottleneck.
Trust is.
Agents can already reason. The next unlock is dynamic identity and permissions at runtime - letting them act safely across the real world.
Building that infrastructure is hard. And essential.
That’s when agents go from demos → infrastructure.
Watch this. 👇 from @ianlivingstone @KeycardAI
Ian Livingstone (@ianlivingstone) CEO of @KeycardAI says the biggest thing standing between today’s AI agents & true autonomy actually isn't intelligence..
“We have apparently AGI, but the real thing that stops us from having autonomous agents that actually work, like a self-driving car, is identity & access.”
“Our systems weren't built for agents that are non-deterministic actors.”
4
4
19
2,743
Must read.
Hard infra used to take years to build and tune. Now an AI agent is helping take a production inference stack from first run to launch in two weeks...with 3.2× the throughput.
Seeing improvements happen this quickly, this deep in the stack, should make you pay attention.
Software infra moats that once seemed safe(r) are rapidly dwindling as well.
We’re sharing how GLM-5.3 helped build and optimize the inference infrastructure serving GLM-5.3-Flash.
The system went from its first successful run to production readiness in less than two weeks, with end-to-end throughput tripling relative to the initial baseline.
The key was dense feedback: local correctness tests, execution traces, microbenchmarks, and end-to-end measurements that enabled targeted hypothesis testing rather than reliance on aggregate performance metrics alone.
z.ai/blog/glm-built-its-infe…
6
6
21
3,686
Jev was two years in the making...and became an “overnight” success.
Amid the insane race from $0 to $100M ARR, remember: hard things take time. Build something amazing and different, and you’ll make your mark.
Plus, is AI starting to build the next AI? 🔥
whatshot.vc/p/whats-in-aiinf…
1
4
27
3,019
Offense has agent swarms. Defense still has humans in a queue.
Agents to keep up with agents. The generator can’t be the validator. Evo by @snyksec is that independent layer — already 60% of new deal volume and growing fast 📈.
snyk.io/news/agentic-ai-secu…
3
1
4
509
this is why Jensen wins no matter what - frontier labs buy GPUs and more open weight models sell more GPUs
we know this is coming - own your intelligence but I'm seeing more and more enterprises stock up on GPUs - cost for sure but also data sovereignty
but also opens up need to secure these systems - going to be a massive market beyond just hyperscalers and neoclouds - enterprises will need all the infra to manage, support and secure
@lavaiodc @spectrocloudinc
Latham & Watkins buys Nvidia servers to set up in-house AI systems ft.trib.al/0CRKbBl
6
7
47
6,547
boldstart ventures retweeted
This is why @Boldstartvc is the best
2
2
6
836
When your startup can be livestreamed in 3 days, starting is the commodity. Enduring is not.
Mitchell Hashimoto nails why: great founders have agency. They see the world as mutable, something they can bend toward what they believe.
And when founders run out of energy, you end up with Bending Spoons buying you...
Keep building the hard, hard thing 🔥
Read the latest What's 🔥 in AI/Infra/VC
whatshot.vc/p/whats-in-aiinf…
6
15
1,338
Big day at @clay!
Doing it the Clay way
$200m ARR now on wya to $24 with mission to double on…
Very little cash burn 👀
More importantly the product velocity is off charts with Act III self learning GTM
Proud investors from Inception 🚀!
Today we're announcing our Series D: $115M raised, led by @Wellington_mgmt, at a $7.1B valuation.
Our mission is to grow every company to its full potential.
To do that, we need to do 3 things ⬇️
4
1
20
2,336
Today we're opening early access to Agent Sessions: #observability for your AI coding agents.
Capture, search, audit, and analyze every @claudeai Code, #Codex, and @opencode session, the way you already do for logs and traces. Support for any @OpenTelemetry #LLM harness is coming soon.
grepr.ai/blog/agent-sessions…
1
2
2
332
Self learning loops for more customers 👀
Many of the fastest growing AI companies use Clay including @AnthropicAI @OpenAI @stripe and many more!
Congrats @clay team!!
Excited for @Boldstartvc to continue investing from Inception!
@Clay just raised a $115M Series D at a $7.1B valuation.
Now, more than 17,000 teams build on Clay, including Anthropic, SpaceXAI, Google, OpenAI, Stripe, Visa, UPS — and 80% of the Forbes AI50.
@nytimes just broke the news, but the story goes far beyond the raise. AI is unleashing the biggest wave of company creation in history. Clay’s goal is to be the engine they use to grow.
Clay is a self-learning revenue engine that figures out the best GTM play for your company, then runs it.
We started as the company with the best data about other companies. Then we became the infrastructure to run campaigns across email, ads, and landing pages.
Now we’re building agents that learn from every campaign, predict the next best action, and get better each time they run.
Software engineers use coding agents to write code. GTM engineers now use growth agents to autonomously grow their companies.
In 2023, we introduced GTME as a new role for people creating revenue systems with data, automation, and AI. Two years later, thousands of GTMEs are growing the world's best companies. They are gathering in nearly 100 Clay Clubs around the world, from Bangalore to Boston and Lahore to Lisbon.
That’s why we’re proud to launch a $1M scholarship fund, helping more people enter the profession.
@Wellington_Mgmt led our Series D, with participation from @sequoia, @stepstonegroup, @a16z Perennial, @MeritechCapital, DST Global, @CapitalG, @BoxGroup, @boldstartvc, @BloombergBeta, and @EvolutionEquity.
A gracious thank you goes out to our customers, community, partners, team, and every GTM engineer shaping this with us.
We are going to unveil the future of Agentic GTM at SCULPT on October 8.
Stay tuned to see how to use AI to grow every company.
🖤🩵❤️💛
2
2
20
1,817
glad more people are waking up to securing AI data centers + neoclouds.
same complexity that took humans decades… agents can walk it in minutes.
rogue agents as tenants, exploring BMCs, GPU fabrics, control planes at machine speed… and the security tools still sit in silos.
@lavaiodc laying it out. read this. 🤯
lavahq.io/research/datacente…
4
9
945