VCs are now sharing screenshots in group chats of Claude discouraging investment in open-source AI infra startups and models.
Obviously there is an absolute EXPLOSION of pitches in inference companies, harness companies, RL-as-a-service companies, open-source tooling currently including Neolabs that plan to open source models as well.
Now the obvious takeaway is: “Claude is biased against open source.” Who cares?
The more unsettling take is: every major AI model has a worldview, and that worldview is becoming embedded in capital allocation.
If Claude’s safety priors cause it to frame open-source AI as dangerous, hard to govern, or less fundable, it’s probably doing the same thing in enterprise buying workflows.
Now investors and executives are obviously smarter, but “influence” just changes which risks get highlighted, which questions buyers should ask, and which vendors are suggested…..
@bnicholehopkins is crushing it in voice evals/simulations and we are so excited to be leading her Series A!!
The first time @sbeechuk and I met her in person, we knew we had to invest. (Even my mom met Brooke once up in the Bay Area and said we should invest - no joke 😂).
Congrats to the entire Coval team!
Overwhelmed by the support on our Series A announcement this morning, including the incredible piece by Chris Metinko for Axios (link in first comment).
The next interface isn't an app. It's a conversation. Every enterprise will have a voice or conversational agent as the front door to their product: handling calls, closing deals, resolving issues, onboarding customers. Autonomously, at scale.
But agents are only as valuable as the trust you can place in them. Coval gives enterprises the infrastructure to simulate, test, and continuously improve their agents so they can deploy with confidence and scale without fear.
We've seen this problem before. The reason we trust autonomous vehicles on public roads is because the evaluation infrastructure got rigorous enough. Billions of simulated miles, continuous monitoring, and relentless iteration before a single real-world deployment. We're building that same foundation for voice and conversational AI.
So thankful to all of our investors for believing in what we're building: Norwest, Swift Ventures, Base10 Partners, Y Combinator, Alumni Ventures, Twilio Ventures, MaC Venture Capital, Fortitude Ventures and many more.
Now, back to building!
@CatNyanpital@isaacinthesky@mil000 Take a look at most successful acquisitions (not PE) over the past decade. This is a fairly normal multiple. It's just the $ price that gets everyone's attention
HR Tech has historically been the unsexy cousin of fintech, devtools, and other app layer companies. Legacy platforms with bad UX, point solutions nobody could get budget for (or wanted to use), and a CHRO who was last in line when the CFO did the annual software review. But there seems to be a profound shift among leaders as it relates to the category.
AI is doing to the people stack what cloud did to finance a decade ago. The companies that crack it will sit on some of the most valuable proprietary data in the enterprise (the kind of structured human performance and skills data that AI model providers are actively seeking out).
Teamed up with @stormventures to write a full market map covering 200+ companies across payroll, talent, analytics, compliance, and the new global workforce layer. If you're building in this space or just have strong opinions on who wins, we want to hear from you. Blog in comments.
It’s still early, but it feels like a new layer is starting to form here. @jleites13 and I put together a map of how this ecosystem is evolving and where we’re seeing companies emerge. Take a look at the blog in the OP
There’s been a lot of talk recently about AI agents paying for things in both consumer and business use cases. It sounds straightforward, until you actually think through what needs to happen for #agenticpayments to work.
We wrote more about it here…
https://t.co/FlmdLaJC0Q
Payments today are still designed around a person clicking a button somewhere. Even if an agent can decide what to do, it still can’t fully execute without running into a very human-shaped system. Rebuilding the infrastructure so they can transact is where the challenge lies.
That’s where the current stack starts to break. Identity, permissions, payments, all of it is still buried in application logic and built around humans being in the loop. Agents need something that’s usable by software. This is where the status quo is beginning to change.
But once you think about it that way, payments stop looking like a separate step and start to feel like part of the workflow itself. If an agent is doing work, it probably needs a seamless rail to pay for things along the way, whether that’s APIs, services, or even other agents.