Clarity under constraint
THRESHOLD SIGNALWORKS
Governance, evaluation, and control for AI systems
AI systems are being deployed without enforceable controls, without evaluation of what they actually know, and without audit trails for when things go wrong. We build the infrastructure layer that fixes that.
Threshold Signalworks builds governance, evaluation, and control infrastructure for AI systems. Keel enforces policy before agent actions complete. Driftwatch measures model behaviour and catches drift before it reaches production. Helmsman corrects inference-time failures without retraining. Together they produce cryptographically verifiable evidence that governance happened.
From coding agents to model evaluation pipelines to production deployments.
Products
Policy enforcement for AI coding agents. Classifies every agent action by risk, requires structured human approval before anything destructive, and logs everything to a tamper-evident audit trail. Constraints live on disk, not in context, so they survive when conversations are compacted. Cloud tier adds independent third-party attestation, policy versioning, and team governance.
Epistemic evaluation and regression tracking for AI models. Measures calibration, drift, overconfidence, and refusal behaviour across model updates, prompt changes, and workflow modifications. Reproducible evaluation runs with full provenance chains. Cloud tier adds cryptographically signed reports, shared baselines, and release gates.
Inference-time stabilisation. Detects when a model is committing to an answer before sufficient information is available, and corrects the behaviour without retraining. Early research validated across three model families with statistically significant hallucination reduction. Feeds calibrated confidence signals into Keel's risk assessment.
Independent trust and governance layer. Syncs policies and audit logs across agents and machines, provides server-signed attestation bundles, and produces compliance-ready evidence exports. The independent third party that verifies your governance actually happened. EU-hosted, GDPR-native.
Research
Threshold Systems is the research arm of Threshold Signalworks. We study how instability enters during inference, tool use, and autonomous workflow execution, and we build measurement and intervention tools grounded in that understanding.
Current work spans AI evaluation protocols, cognitive architecture under constraint, and human decision-making in high-uncertainty environments. Publications and artefact packs are released through threshold.systems.
Artefacts
Public artefact packs (evaluation runs, reports, provenance chains) will appear here as they are released.