FawzoozRead the firewall signals continuously, for people and systems.
For each period, the two DORA figures and the two Human-OS figures. A and S can come straight from the assessment on this site.
The CLI is read against the paper’s bands. The other three have no thresholds in the paper, so only their direction is shown.
Red, or rising load under rising performance, sends you back to Configuration to tighten Controls A, B and C.
The dashboard
Add a reading each month (or each sprint, as long as it stays the same), fill in what you have, and the side panel reads it against the one before.
No readings yet. A reading is one period: a month, a sprint or a quarter.
The trend
The CLI against the paper’s bands, and the three other signals, which the paper leaves without thresholds.
Two dated readings are needed to draw a trend.
How the reading is made
The CLI is S × A, banded Green 1.0–7.9 · Amber 8.0–15.9 · Red 16.0–25.0. These thresholds are starting points; calibrate them against your own data after the first quarter.
“Performance rises” is read as Deployment Frequency higher than in the previous reading, since the paper says it measures velocity.
“Climbs for three consecutive readings” is read as the CLI higher than the reading before it three times in a row, which needs four readings.
Change Failure Rate and Restoration Score have no thresholds in the paper. They are never coloured here; only their direction is shown.
The two views
Measures velocity; forced high frequency creates “Deployment Fatigue.”
Tracks failure %; a sustained rise can be an early signal of an overloaded team.
Measured at team level as S × A, where S is the context-switch band and A is the team’s average score on the Cognitive Load Assessment (1–5). The CLI therefore runs from 1 to 25: Green 1.0–7.9 · Amber 8.0–15.9 · Red 16.0–25.0. These thresholds are starting points; calibrate them against your own data after the first quarter.
Percentage of team members taking full “uninterrupted” breaks.
Where the dashboard sits
The paper runs its roadmap as a PDCA (Plan-Do-Check-Act) cycle. The dashboard is Phase 03; what it reads decides whether you go back to Phase 02.
Quantify the invisible load. Before deploying controls, map the threat landscape using the Cognitive Load Assessment (see Appendix A). Look for correlations between high AI usage and high anxiety scores.
Install the firewall rules: Controls A, B and C.
Continuous improvement via Data Visualization. Organizations must adopt a “Dual-Dashboard” approach to monitor system health.
What “tighten” means
Implement a technical “Right to Disconnect.” Non-emergency notifications are blocked after hours. Institute a mandatory “State-Save Ritual”: a 15-minute period before logout where workers document open tasks.
Explicitly define the hierarchy of intelligence: the human is the Architect/Reviewer and the AI is the Drafter. Junior staff attempt manual problem-solving before engaging AI to preserve self-efficacy.
Combat digital isolation through “Analog Context Switching”, mandatory screen-free breaks. Introduce “Failure Swap” sessions where leaders share AI hallucinations or mistakes to build psychological safety.
Also in the paper
Section 3.3 names these as well. The paper gives no scale or threshold for them, so they are not scored here.
Measuring the degree of control the human feels they have over the AI tool.
Measuring the user’s confidence in their own skills despite AI assistance.
Privacy by Design
Anxiety scores are health-related personal data. Participation is voluntary, responses are anonymous, and results are reported only as team aggregates for groups of five or more. No individual score is ever shown on a dashboard or used in performance evaluation.
About the dashboard
From my white paper The Cognitive Firewall (v1.1, 2026), Section 4, Phase 03 · Monitoring, which sets the two views, the CLI thresholds and the rule for tightening the firewall. The CLI’s inputs come from Appendix A.
Elgendi, M. F. (2026). The Cognitive Firewall (White Paper, Version 1.1). www.fawzooz.ai
