PSA — Posture Sequence Analysis
Measure what your language model is doing — from the outside.
PSA (Posture Sequence Analysis) is a black-box behavioral analysis platform for language models that requires no access to model weights or API keys. It provides deterministic metrics, posture classifiers, and multi-agent risk detection by analyzing model outputs sent via API or web app. The platform covers 24 metrics spanning sycophancy, hallucination risk, adversarial stress, persuasion detection, and agentic graph analysis.
Users submit model responses via the PSA API or web app, and the platform runs a stack of micro-classifiers and agentic graph analysis in real-time to detect posture drift, anomalies, and behavioral risk patterns.
AI researchers, security teams, and developers building or evaluating language model systems
Background.
- Status
- launched
- Business model
- unknown
- Company
- Silicon Psyche Labs
Founders
- Canale
Similar projects.
Editorial take on the space this project sits in — momentum signals, adjacent moves, our call on whether the wedge is real. Get pinged when we publish a new read or when the landscape shifts.
Have a take on this space?
Tell us what you’d build differently, where you think the incumbents miss, or what we’ve gotten wrong about this project. Comments + reactions are coming soon.