Meet Argus: Your Always-On Digital Guardian, Now Self-Serve

We have been calling it 'the Protector Agent' in internal documentation for approximately four months now, which is about three months and twenty-nine days past the point when it needed a better name. As of today, its name is Argus.

Argus, for the mythologically inclined, was the hundred-eyed giant of Greek mythology — the all-seeing watchman who, inconveniently for the myths, never slept. We find this extremely on-brand for a product that literally never sleeps and monitors with figuratively many eyes. Our lawyers asked us to clarify that Argus has zero actual eyes and is entirely software.

Argus Is Now Self-Serve

When we first shipped the Protector Agent in closed beta, every deployment required a manual configuration process we diplomatically referred to as 'assisted onboarding' and less diplomatically referred to as 'a lot of work.' That era is over. Any CleverThis account can now spin up their own Argus instance without contacting us, without a consultation call, and without waiting for us to do it for them.

Built on GISM-0: Reasoning That Shows Its Work

Argus is the first self-serve product built on GISM-0, our Graph-Native Inferencing Semantic Model. Where most monitoring tools apply static thresholds or opaque anomaly scores, Argus reasons over a live concept graph of your systems, accounts, and activity patterns. Every alert is fully explainable — you can see exactly which concepts, relationships, and graph paths triggered it.

“When something wakes you up at 3am, 'anomaly score: 0.94' is not an adequate explanation. 'Your auth service accepted tokens from three IP ranges that have never appeared in your system graph' is.”

Standalone GISM-0 self-serve — independent of Argus, available as a general-purpose graph-native reasoning API — is on track for the next update. If you're building something that needs explainable, relationship-aware reasoning without wrapping it in a specific agent product, that's the one.

Platform-Wide Subscription Endpoints

This month we also shipped something that quietly makes a lot of other things possible. CleverThis now supports custom container-backed endpoints — meaning you can use a Claude Max subscription as the inference backend for any CleverThis endpoint, not just Argus. These endpoints expose the same OpenAI-compatible API surface as the rest of the CleverThis gateway. Your existing integrations keep working. You just swap the backing provider.

CleverLang: The Protocol Under the Platform

We want to be direct about CleverLang before anyone describes it as a product line in a pitch deck: CleverLang is not a product. It is not a thing you purchase. It is a semi-structured language standard for technical communication — designed for AI-to-AI and AI-to-human exchange of code structure, architecture decisions, and system state.

“You don't buy JSON. You don't subscribe to markdown. CleverLang is that kind of thing — an open protocol standard that lives underneath many of the products you will build. It ships as open specification and belongs to the commons.”

CleverLang is the communication layer that makes AI agent handoffs coherent, that makes an AI's description of what it just did reviewable by a human without losing technical fidelity, and that makes code-level communication consistent across the CleverThis platform. It is everywhere in what we're building.

What's Coming Next

We remain a small team moving, by our own perhaps optimistic self-assessment, extremely quickly. The commit frequency has not decreased. The ambition has also not decreased, which is either encouraging or concerning depending on your perspective. We choose encouraging.