


Co-Founder & Partner @NextViewVC Seed Investor: @attentivemobile @TripleLiftHQ @thredUP @ParsecTeam AI is disrupting VC: carriedawayvc.substack.com


So the diligence process gets shorter, and it can shrink a long way before it gets any less rigorous. The hard conversations stay, and so does a night to sleep on it. Most of the rest was theater, work done so it can be seen. carriedawayvc.substack.com/p…

The real critical thinking about an investment takes an hour or two. The document exists so we can say we did the work. Some of that thinking may end up the machine's too. The audience changes first: a partner who gets the packet overnight knows how little it proves.

AI has already made VC desk work nearly instant: the competitive map, the product comparison, the exit multiples on the comps. The memo and the IC deck are next. A memo written by a machine forces a firm to ask how much thinking ever happened in the writing.

Proud to back @boxerbk and Chris a second time with @Phaze_in. Years ago my memo for their previous startup Parsec said they were “shifting GPUs from local hardware to servers in the cloud.” Same idea, much bigger moment, now that agents need computers too. Why we @NextViewVC invested with them again: nextview.vc/blog/why-we-at-n…
Today we’re announcing an $8.75M seed round for Phaze. Companies aren’t ready for how many computers people and AI agents are about to need. And it’s happening just as computer prices are up 3x. Chris and I spent years building Parsec to make a computer thousands of miles away feel like it’s sitting right in front of you. We sold that company, and we’re starting over because people aren’t the only users of computers anymore. Now, agents need access to real computers to run applications, test software, and get work done. They need to work alongside people, hand off control when necessary, and operate within the security policies companies already have in place. And companies need to start managing computers like infrastructure. That’s why we’re building @Phaze_in: secure, high-performance access to computers, wherever those machines happen to be. An engineer can connect to a CAD workstation from any device and forget about being physically stuck at that workstation. A video editor can ask an AI agent to connect to a render machine to monitor and fix broken renders. An IT support person can fix multiple machines at the same time with an AI agent connecting to employees’ computers. If the agent needs help, it can hand control back to the technician in the same session. We loved working with the firms who backed Parsec, so we’re running it back. Our Seed round is led by @martin_casado at @a16z, with @bryce at @indievc, @davidbeisel at @NextViewVC, @BenjLerer at @LererHippeau, and @Ryann_Lai at @makersfundvc.

Now the public record, read at the scale only AI can manage can replace the typical introduction. A trusted endorsement still opens the door. But most of us haven't decided whether we'd back a founder with no one we know vouching for her. carriedawayvc.substack.com/p…

The intro was a rationing device. Up to a couple hundred of first meetings a year per partner, cut along the VC's own network: two handshakes away gets in, four away doesn't. This year our system flagged a founder nobody had introduced, read her team's record before our first call. We invested.

In 2013 I wrote that a warm intro was neither necessary nor sufficient to get a VC meeting. Then my firm @nextviewvc spent thirteen years backing almost nobody who came without an introduction. That tells you what our filter let through. It can't tell you what it missed.

The team at @NextViewVC 's portfolio company @skyfallai has just introduced the first Engineering World Model, beating existing frontier models in CAD and BIM designs with higher precision while costing less.
Introducing SKY-001, the first Engineering World Model. On CAD and BIM tasks, it beats GPT-6 Astra and every frontier model we tested: -> 51-125x faster - > 150-210x cheaper -> Higher IOU than all frontier models SKY-001 lets you dream a physical object before you build it. 18 months of work. Built by the team out of Maluuba's deep learning lab (acq. by @MSRResearch). The first ‘AI General Engineer’ will be the backbone for Autonomous Manufacturing.


For twenty years, finding founders took a network and meeting them took hours. Now data and software read the entire legible field at once. Venture was paid to find, judge, and be let in. Sourcing was the first defense, and the first one AI takes away. carriedawayvc.substack.com/p…

In 2015 a founder stood in the room at one of our own events, flagged to me by a colleague. I walked past the handshake. His company is worth billions and I've still never met him. Not a near miss. A near miss needs a meeting. What failed wasn't judgment. It was coverage.

Most VCs are aiming at a bullseye on a target they can barely see. Last year the venture industry did about 15,000 deals. A typical early-stage firm backed twelve of them. The one that cost it most is the one it never saw.

A year later and the seed VC crisis has only sharpened. My partner @robgo at @NextViewVC lays out the three paths forward, and the harder truth that most firms are committing to none of them. Essential read for anyone in early-stage.
It's been about a year since I wrote my post about the Existential Crisis for Seed VC's. Some updated thoughts here on the three paths forward for seed investors. nextview.vc/blog/the-three-p…

As part of testing @Muse, I quizzed it to explain to me a few things I think about when using these agents and deciding how much connections and context to give - the struggle of the Trust-Utility Curve (the more you give, the more useful it can be). x.com/melodykoh/status/20262… TLDR - there continues to be limitations around how LLMs fundamentally work and the probabilistic nature of things, and there really isn't a way to get around on context leakage or prompt injection-like attacks. However, I found Muse/Meta's willingness to be transparent in its mechanics reassuring. A few highlights on my Q&A with Muse - see more details on the screenshots of my questions and its responses. 🧵 Q1 - you say that you will ask for my permission for actions, but how do I know this can be enforced deterministically (vs instructions in files, which are just vibes for these models)? Most reassuring part of the answer: "When I invoke a send tool, the runtime intercepts the call and blocks execution until you've approved it. I can't trigger the approval myself, can't see it, can't override it — the approval card is invisible to me and your decision is final. So even a compromised or injected version of me, actively trying to send without asking, hits the same wall: there is no code path where the send executes without the runtime obtaining your decision first." Translate to non-developer speak: there's a mechanical gate that forces the model to serve up approval to the user when the action is a list of predefined "must get user approval". No vibes-bsed (good!)
Finally caved in to try Muse. Quick hits / thoughts thus far (especially compared to Instinct)- 1. I really like the ability to see the detailed actions (tool calls, steps) behind the scene - they do a good job hiding these but they’re also easily accessible and not completely black-box / headless. Why is it useful? It’s a level of transparency and also lets me intercept if I see the agent going down the wrong path E.g. I asked it to draft something for me and it found a GitHub public repo on content by itself. I saw it (good job finding it!) and had to intercept and tell it that the content there is few months behind my private repo, so don’t use that as authority 2. The app interface allows for much more robust interaction modalities. I especially like how you can see browser use actions (like literally what it’s doing on the browser) and you can take over at any time. In contrast, having used Instinct for a few weeks now, the single-threading iMessage interface is a bit unwieldy and hard to manage, especially when having it do multiple tasks and needing to find stuff from earlier conversations More to come. Only 30mins in!

The first stock I ever bought was a chain of giant sneaker superstores called Just For Feet. And I lost every dollar.
