You know who is using AI. Do you know how they operate it?
MO§ES™ measures how people operate AI systems across real tasks, workflows, models, and operating conditions using content-free token telemetry. Not usage analytics. Not a knowledge test. Not self-reported proficiency. Nonlinear intelligence from linear telemetry; in 30 days.
Book a Demo Run the DemoUpsilon — Enterprise AI Operating Intelligence
Establish a 30-day baseline of how your people and AI systems operate together.
Upsilon measures two sides of the enterprise AI operating system: operator efficacy and system intelligence. The result is a practical baseline of the operator × system relationship across real workflows, models, and tools.
Operator Efficacy
How effectively people operate AI systems across real workflows, tools, and operating conditions.
System Intelligence
How models, tools, agents, and workflow environments support or constrain effective operation.
The Missing Operating Layer
Adoption is not efficacy. Usage is not system intelligence.
Enterprise analytics can show who has access, which tools are being used, and how often. That does not tell you where operating capability is strong, where it is concentrated, or whether friction comes from the operator, the system, or the workflow.
Operator
Where is efficacy strong, unstable, concentrated, or changing?
System
Which models, tools, or agents improve or degrade the operating relationship?
Workflow
Where does friction originate; and what should be changed first?
From Usage Data to Operating Intelligence
You already collect usage data. Upsilon shows you what the composition means.
Most enterprise AI programs already collect operational endpoints: licenses, active users, message counts, spend, model usage, tool adoption, and completion activity. Upsilon uses that existing usage and operational data as raw material, then applies compositional analysis to reveal how operators, systems, tools, models, and workflows are actually working together.
1. Existing Data
Usage, model, tool, session, token, workflow, cost, and other operational signals already produced by the environment.
2. Compositional Analysis
Measure the structure and relationship between those signals rather than treating each endpoint as an isolated activity count.
3. New Business Insight
See operator efficacy, system intelligence, fit, friction, concentration, adaptation, and implementation differences that usage dashboards do not expose.
What the New Data Layer Reveals
Insights companies typically cannot get from adoption metrics alone.
Operator Baseline
How are different people actually operating AI?
Capability Distribution
Is capability broadly distributed or concentrated in a small number of operators?
Operator × System Fit
Does the same operator perform differently across models, tools, or environments?
System Intelligence Profile
Which systems produce stronger operating relationships with this workforce?
Workflow Friction
Is the bottleneck the operator, model, tool, agent, or workflow?
Adaptation Over Time
Are operators changing how they work as they gain experience?
Implementation Comparison
Did Tool B, Model B, or a new workflow materially change operating behavior?
Training / Intervention Measurement
Did the intervention change the operating structure or merely increase usage?
Concentration Risk
Does the organization depend disproportionately on a handful of strong operators?
Cohort Differences
Are teams, roles, workflows, or operating environments behaving differently?
From Insight to Implementation Decisions
Move beyond "AI adoption increased."
Upsilon gives the organization a baseline against which different implementations can be compared. The question is no longer only whether AI usage increased, but what actually changed in the operating system.
Baseline
Who are our operators? How do they currently work? How does each system perform with them?
Implementation Comparison
Model A vs Model B. Tool A vs Tool B. Agent workflow vs manual workflow. Training before vs after. Workflow design A vs B.
Re-measure
What changed? Who improved? Which system improved? Where did friction move? Did capability spread or become more concentrated?
Decision
Use better evidence for AI deployment, tool selection, training, workflow design, system configuration, and future investment.
The 30-Day Pilot
Baseline the operating system. Compare implementations. Measure what changed.
The pilot creates the reference state needed to evaluate future changes. It establishes the current human × AI operating system, creates comparative intelligence across operators, systems, tools, workflows, teams, and implementations, and gives the organization a repeatable measurement loop.
1. Establish the Baseline
Measure the current operator and system state across selected workflows and cohorts.
2. Compare Implementations
Compare models, tools, workflows, teams, training states, or system configurations against the baseline.
3. Create the Measurement Loop
Change something → measure again → determine whether the operating system actually changed.
What You Know in 30 Days
A measured baseline for your enterprise AI operating system.
Upsilon establishes a starting state, identifies where differences originate, and creates a framework for targeted intervention and re-measurement.
- Operator and cohort baselines
- System and model intelligence profiles
- Operator × system fit analysis
- Capability concentration analysis
- Workflow friction identification
- Longitudinal adaptation signals
- Intervention targets
- Re-measurement framework
How the Measurement Builds
Start at the operating event. Build upward.
The enterprise picture is built from measured operating relationships rather than a score assigned from the top.
From Baseline to Action
EVALUATE → BENCHMARK → DIAGNOSE → INTERVENE → RE-EVALUATE.
01 Evaluate
Establish the current operating state across operators, systems, workflows, and time.
02 Benchmark
Compare operators, systems, and teams against the cohort reference field.
03 Diagnose
Separate operator effects, system effects, workflow friction, and concentration patterns.
04 Intervene
Change training, tooling, workflow design, system selection, or operating conditions deliberately.
05 Re-evaluate
Determine what changed and whether the operating system improved.
Demo Provenance
Observed data. Simulated enterprise environment.
The demonstration uses real operator measurements sourced from the SignalAF reference field. The enterprise cohort structure, timeline, workflow scenarios, and intervention sequences are simulated for demonstration purposes. The underlying operator measurements are observed data.
Underlying Architecture
One commercial offer. Distinct underlying systems.
SignalAF
Measurement infrastructure. Observes how a human operates a technological system.
SigRank
Evaluation and reference field. Locates measured operator behavior against a wider field.
Upsilon
Commercial enterprise layer. Packages the 30-day baseline, diagnosis, intervention targets, and re-measurement.
MO§ES™
Underlying architecture. Extends the system beyond measurement into governed execution and continuity.
Four Questions
You can't answer these with usage analytics.
Your observability tools tell you who has access, how many tokens, how often. Month after month, the same linear data. These questions require something else.
Do you know which AI systems your staff produces more with?
Not which ones they use most. Which ones they actually produce more with.
Is changing software a blind roll of the dice?
You're about to switch models, tools, or platforms. Will it help, hurt, or do nothing? You don't know.
Do you know who your tech leaders are?
Not by title. By operating capability. Who makes the system work better by being in the room?
Do you have an AI savant in your company; and have no idea?
Someone whose operating relationship with AI is quietly outperforming everyone else. Your dashboards don't show that.
Why Usage Analytics Isn't Enough
Linear data gets linear results.
A thermometer tells you the temperature. But weather isn't a temperature. It's a nonlinear system of pressure gradients, humidity, wind shear, and jet streams. You can't understand weather by reading a thermometer. You can't understand operating intelligence by counting tokens.
Linear: What you have now
Token counts. Active users. Spend. Who has access. How often. The same endpoints, month after month. Too many tokens, no tokens, more tokens, fewer tokens. Each signal in isolation.
Nonlinear: What Upsilon reveals
The relationships between the signals. How operators and systems compose. Where capability concentrates. Where friction originates. What changes when you change something. Intelligence that doesn't exist in any single metric. It emerges from the composition.
30-Day Enterprise Baseline
Know how your people and AI systems operate before deciding what to change.
Start with a bounded team, workflow, model set, or tool deployment. Establish the baseline, identify where differences originate, and leave with a framework for targeted intervention and re-measurement. Bespoke evals, analysis, and solutions for enterprises.
Go deeper.
Concept definitions, how-to guides, competitor comparisons, and frequently asked questions.
The eight Upsilon metrics (Yield, Leverage, SNR, Velocity, 10xDEV, Scale V, Construction, Efficiency) and five framework definitions (Divergence, Benchmark, Intervention, Canonical Telemetry, Governance).
What AI evaluation means, the four layers (model, output, safety, operator), and why the operator layer is where ROI lives or dies.
How to evaluate AI operators, how to measure performance, eval vs usage analytics, eval vs skills assessment.
Frequently asked questions about AI operator evaluation, measurement, governance, and the pilot process.
How MO§ES™ compares to Workera, Worklytics, Weave, Paxel, Vals AI, Bryq, Canditech, genAssess, AI Acumen, and Prompt Ranks.
Best AI operator evaluation tools, Workera alternatives, Worklytics alternatives, AI workforce measurement tools, AI skills assessment alternatives.
The full evaluation framework: five questions, canonical telemetry, derived metrics, percentile bands, intervention testing.
Related systems.
Public leaderboard and benchmark for AI operators. signalaf.com
Enterprise measurement engine. mos2es.org
Dual-governance agentic marketplace. signomy.xyz
Application capital from previous work. mos2es.xyz