@heyshrutimishra: The real moat was never the product. It was the infrastructure that turned users into revenue. Most founders skip this …

X AI KOLs Following Tools

Summary

Velobase open-sources its entire infrastructure stack, arguing that the real moat for startups is not the product but the revenue-generating infrastructure behind it.

The real moat was never the product. It was the infrastructure that turned users into revenue. Most founders skip this layer entirely. This post is a masterclass in what they’re missing.
Original Article
View Cached Full Text

Cached at: 05/20/26, 06:26 AM

The real moat was never the product.

It was the infrastructure that turned users into revenue.

Most founders skip this layer entirely.

This post is a masterclass in what they’re missing.

We’ve been watching the wrong AI story.

While the timeline keeps debating whether Mythos is real, hyped, or just well-marketed, 360 has quietly taken an autonomous vulnerability-mining agent into the OpenClaw ecosystem and walked out with 23 confirmed vulnerabilities, including two Criticals, all filed into the CNNVD and CNVD national vulnerability databases.

That is the gap most of the timeline is stepping over.

The OpenClaw work is also closer to where AI security is actually heading. Agents are a fuzzy-boundary target where prompts double as instructions, where the static-analysis baseline barely exists, and where a breach at any one of authentication, network, execution, or control cascades through the rest. The 23 findings cover all four layers.

Where the Mythos conversation is still about whether the capability is real, the 360 work shows the same capability already running as operating infrastructure, on a target that better reflects the next phase of AI security.

The interesting question stopped being “is AI-on-AI security real.” It is who owns enough attacker-perspective data to make it production-grade.

Similar Articles

Moats Need Models (6 minute read)

TLDR AI

The article argues that AI defensibility comes from owning the full feedback loop—custom models post-trained on proprietary data, tuned to specific workflows, and evaluated by user-defined standards—rather than renting frontier APIs from suppliers who can change terms. It emphasizes model customization as key to differentiation and margin control.