Methodology

How We Measure AI Training Outcomes

The SIRA 7-layer framework, the Medha Grade, and the verification model that prove whether AI training improved productivity, reduced dependency risk, and created measurable value.

The Framework

Seven Layers of AI Risk

SIRA measures AI adoption risk across 7 interconnected layers — from energy costs and infrastructure outages, through vendor concentration and application errors, to workforce displacement and employee deskilling. This is what changes when you train teams on AI.

L7
HumanMost critical, least examined

Are your people losing the ability to work without AI? Deskilling, over-reliance, eroding judgement.

L6
Workforce

Are jobs being cut or hollowed out? Layoffs, role displacement, lost institutional knowledge.

L5
Market

Is AI hype moving your stock price or distorting your competitive landscape?

L4
Vendor

Could you switch AI providers tomorrow? Lock-in, concentration, single points of failure.

L3
Application

Is the AI giving wrong answers? Hallucination, data leakage, integration failures.

L2
Infrastructure

What happens when the cloud goes down? Outages, compute dependency, latency.

L1
Energy

How much power is your AI burning? Data center costs, sustainability, resource strain.

Each instrument is assessed across all seven layers with proprietary sensitivity weights. High-exposure layers drive the grade; low-exposure layers are monitored but weighted accordingly.

The Grade

A Credit Rating for AI Health

Seven grades from AAA (AI helps, no dependency) to CCC (AI destroying value). Your grade reflects how dependent your people are on AI, whether you could switch tools, whether key roles are being displaced, and how much a vendor outage would cost. Train your team — watch the grade improve.

Medha-AAAAI helps, no dependencyYour team uses AI effectively and can work without it. Best possible position.
Medha-AAHealthyMinor dependency risks, but AI is clearly creating more value than it costs.
Medha-AWatch closelyEmerging risks in 1–2 areas. Worth monitoring before they grow.
Medha-BBBAction neededMeasurable problems. Time to restructure how your team uses AI.
Medha-BBStressedMultiple problem areas. Intervention required — training or tool changes.
Medha-BCriticalAI is costing more than it delivers. Scale back non-essential AI usage.
Medha-CCCAI destroying valueAI is actively hurting productivity, creating dependency, or displacing key people.

Grades are derived from proprietary scoring of event severity, layer exposure weights, and structural risk indicators. The specific thresholds and calibration model are internal to Purna Medha.

Value at Risk

What Could This Cost You?

We calculate two kinds of financial risk from AI adoption — what you're wasting on tools, and what it costs when people get displaced. Both are priced in your local currency.

AI Spend Risk

How much of your AI tool spending is being wasted? This measures the portion of your annual AI investment at risk of producing no value — based on your Medha Grade and sector benchmarks.

Converted to local currency at indicative exchange rates.

Workforce Disruption Cost

What does it cost when your people get displaced or deskilled? This is priced at local salaries — because losing a worker in Mumbai costs differently than in Zurich. Not a currency conversion; a real local calculation.

Displacement factors are proprietary, derived from client data and reversal research.

Why location changes the number

The same Medha Grade in India produces a very different cost than in Switzerland. AI tool spending is similar globally, but workforce displacement costs vary 5-10x based on local salaries. Your risk number reflects your actual market — not a generic global average.

The Missing Layer

Enterprise Risk + Intelligence Risk

Most AI risk frameworks only measure what AI does to your business. We also measure what AI does to your people's ability to think. These are two different risks — and they compound each other.

Enterprise Risk

What AI does to your business. Vendor lock-in, outage costs, workforce displacement, wasted tool spend. Measured per sector with the SIRA 7-layer framework. This is the Medha Grade.

Assess your organisation

Intelligence Risk

What AI does to your thinking. Are you building real capability or just dependency? Most people use AI in ways that make them faster today and more fragile tomorrow. This is the layer nobody measures — until now.

Assess yourself

Three AI Operating Levels

How you use AI determines your intelligence risk. Most people are at Level 1. They don't know it.

Level 1AI UserHigh intelligence risk

Uses AI as a faster search engine or writing assistant. Copy-paste productivity. Feels faster, changes nothing structurally. If the tool disappears, you lose convenience — but you also haven't built anything lasting.

90% of AI value untapped. High dependency on a single tool. No verification. No measurement.

Level 2AI OperatorMedium intelligence risk

Multiple tools, real speed gains, genuine fluency. But the tools run in parallel, not in concert. You're doing more of the same things faster — you haven't changed what you do or how you think.

Can't prove the productivity gain. No verification cost measurement. Doesn't know dependency risk.

Level 3AI ArchitectLow intelligence risk

AI has changed how you approach problems. You chain tools, build workflows, verify output. AI is infrastructure, not a feature. But even here — can you measure your verification cost? Is it going up or down?

Flying blind at high speed. Knows AI works but can't prove it with a number.

Why this compounds enterprise risk

Your enterprise might have a healthy Medha Grade — but if 80% of your people are at Level 1, one tool change collapses everything. Intelligence risk is the hidden multiplier on enterprise risk. A team of Level 1 users with Copilot is more fragile than a team of Level 3 architects with no AI tools at all. We measure both.

Instruments

10 Sectors We Track

Each sector of AI adoption gets its own grade and risk number. Different industries face different risks — a healthcare AI failure looks nothing like a developer tool outage.

GLBL.AI

Global AI Ecosystem

Aggregate risk across all sectors

SAAS.AI

Enterprise SaaS AI

Copilot, Gemini Workspace, Claude for Work adoption

FIN.AI

Financial Services AI

Trading, risk modelling, fraud detection, client servicing

MED.AI

Medical & Healthcare AI

Clinical decision support, diagnostic imaging, drug discovery, surgical robotics

CXOP.AI

Customer Operations AI

Support/success teams replaced or augmented by AI agents

AUTO.AI

Autonomous Systems

Self-driving, robotics, drone delivery

DEVX.AI

Developer Productivity AI

Coding copilots, AI-assisted CI/CD, testing

LAW.AI

Legal & Compliance AI

Contract review, legal research, due diligence, compliance monitoring, e-discovery

IND.AI

Heavy Industry & Manufacturing AI

Predictive maintenance, quality inspection, supply chain optimisation, digital twins, industrial robotics

GAME.AI

Entertainment & Gaming AI

AI-generated content, procedural worlds, NPC intelligence, real-time rendering, training simulation

Each instrument has proprietary layer sensitivity weights and an AI spend benchmark calibrated from enterprise data. New instruments are added as AI adoption expands into new sectors.

Layer Sensitivity

How We Weight Each Layer per Instrument

Each instrument's SIRA layer sensitivity controls the heatmap and determines which events affect which instruments most. Weights range from 0 (no exposure) to 1.0 (maximum exposure).

Domain Relevance

Which SIRA layers are structurally important to this sector? For MED.AI, L3 (Application) is 1.0 because a diagnostic hallucination IS a misdiagnosis. For IND.AI, L2 (Infrastructure) is 0.9 because factory uptime depends on cloud and compute reliability.

Failure Consequence

What happens when this layer fails for this instrument? L2 outages in healthcare (MED.AI) are life-critical (weight 0.8), while L2 outages for developer tools (DEVX.AI) are inconvenient but not dangerous (weight 0.4).

Historical Event Correlation

Which layers have historically produced the most impactful events for this sector? CXOP.AI weights L7 (Human) and L6 (Workforce) at 1.0 because the Klarna pattern — mass replacement followed by reversal — plays out primarily in those layers.

Regulatory & Liability Exposure

Which layers carry regulatory risk? FIN.AI weights L3 (Application) at 0.9 because hallucination in client-facing financial tools triggers regulatory liability. LAW.AI weights L3 at 1.0 because fabricated citations in court filings have already caused sanctions.

Dependency Depth

How deeply does this sector depend on AI at each layer? AUTO.AI weights L1 (Energy) at 0.5 — higher than any other instrument — because autonomous systems require sustained compute power with zero interruption.

Calibration Process

Layer weights are calibrated through a three-step process:

  1. Initial assignment based on sector domain analysis and structural dependency mapping.
  2. Back-testing against historical events — do high-weight layers correspond to the layers where real incidents occurred?
  3. Client validation — weights are refined as enterprise assessments reveal actual layer exposure in the field.

The specific weight values are proprietary. The methodology is transparent; the calibration numbers are internal to Purna Medha.

Central Benchmark

How Fragile Is Your AI Setup?

The Medha Grade tells you how risky things are now. The SPE tells you how much shock your organisation can absorb. Can you survive an AI outage? Can you adopt a new tool without breaking everything?

Stability — Can you survive without AI?

If your AI tools disappeared tomorrow, could your team still function? High stability means human processes still exist as fallbacks. Low stability means operations would collapse.

Plasticity — Can you adopt new AI safely?

When the next AI tool comes along, can your team integrate it without creating fragility? High plasticity means safe adoption. Low plasticity means each new tool adds compounding risk.

70–100%ResilientCan function with or without AI. Augmentation model, not replacement.
40–69%ModeratePartial AI dependency. Some workflows need AI, others have fallbacks.
0–39%BrittleHigh AI dependency, low rollback capacity. Removal causes breakage.

Why SPE matters

The Medha Grade tells you how much risk exists now. The SPE tells you how fragile the system is if conditions change. A Medha-BBB instrument with 25% SPE is far more dangerous than a Medha-BBB instrument with 60% SPE — because the first one cannot absorb shocks, while the second can fall back to human processes. SPE is the difference between a controlled correction and a systemic failure.

Infrastructure Risk

How Exposed Is Your Industry?

Some industries depend heavily on AI infrastructure — GPUs, cloud, energy. If there’s a chip shortage or a cloud outage, how hard does your sector get hit? The SEI measures this industry-level exposure separately from your company-level risk.

Think of it like electricity

You don’t pay for coal or turbines — you pay an electricity bill. But if the power grid fails, everything stops. AI works the same way. Your company pays for Copilot or ChatGPT subscriptions, but underneath:

GPU chipsNvidia / TSMC
Cloud computeAWS / Azure / GCP
AI APIsOpenAI / Anthropic
Your subscription$30/seat/month

If TSMC (makes 90% of advanced chips) has a production issue, the shock ripples through every layer to your team. You don’t see the risk in your monthly bill — but it’s there. That’s what SEI measures.

What determines your industry’s SEI score

40%
How much compute do you need?

A factory running real-time quality inspection (IND.AI) needs constant cloud compute. A law firm doing document review (LAW.AI) needs much less.

25%
How much energy does your AI burn?

Data centers are competing for grid capacity. Industries running large models or real-time inference are more exposed to energy constraints.

35%
Are you locked into monopoly suppliers?

If your sector depends on a single chip maker, a single cloud provider, or a single API — you’re concentrated. Concentrated = fragile.

70–100%CriticalDeep dependency on monopoly infrastructure. Supply shock cascades immediately.
50–69%ElevatedSignificant exposure. Cloud abstracts but doesn’t eliminate concentration.
30–49%ModerateSome infrastructure dependency. Lighter compute needs reduce exposure.
0–29%LowMinimal infrastructure dependency. Operations function without GPU-intensive compute.

What happens when infrastructure breaks

High-SEI industries don’t just lose more — they lose disproportionately more. A chip shortage that barely affects a law firm can multiply a manufacturer’s risk by 3.65x. Here’s what that looks like across sectors:

IND.AI92%Critical3.65x
AUTO.AI75%Critical2.95x
GLBL.AI66%Elevated2.61x
MED.AI50%Elevated2.06x
DEVX.AI42%Moderate1.82x
LAW.AI32%Moderate1.54x

The bottom line

Imagine two companies with the exact same Medha Grade and team size. One is a steel manufacturer. The other is a law firm. A global chip shortage hits. The manufacturer’s actual risk is 3.65x what the law firm faces — because its AI runs on real-time compute that can’t wait. Your industry determines your infrastructure risk more than your company size does. Most AI risk frameworks ignore this. The SEI makes it visible.

The Core Claim

We Warn Before It Happens

A risk framework is only valuable if it warns you before problems hit. We track structural indicators that consistently show up before the actual events.

The value of risk assessment = Δt between warning and event.

If Δt > 0, the assessment has value.

Safety departures predict governance failures. Vendor concentration predicts outage impact. Workforce displacement patterns predict reversal rates. The SIRA framework tracks these structural indicators and measures the time between signal and event across all seven layers.

Specific timing proof data — which predictions were made and when they materialised — is tracked internally and shared with audit clients. The Pulse page shows events as they occur; the timing proof validates the framework over time.

Severity

How Events Are Scored

Each event on the Pulse is classified by type, assigned to a SIRA layer, and scored for severity on a 1–5 scale.

Safety Exit

Resignation or termination of AI safety personnel

Outage

Service disruptions at major AI vendors

Market

Stock/sector impact from AI-related events

Layoff

AI-driven workforce reductions

Incident

Physical harm, data breach, or financial loss from AI

Reversal

Companies reversing AI displacement decisions

1–2StableNormal conditions, isolated minor events
2–3ElevatedNotable events in 1–2 SIRA layers
3–4StressedMultiple high-severity events, action needed
4–5CriticalSystemic signals, immediate assessment recommended

The severity scoring rubric — how each event type maps to a specific score — is proprietary. Events are curated and scored by the Purna Medha team.

Data Sources

Where the Signals Come From

Eight categories of data feed the Pulse, each mapped to specific SIRA layers.

AI Incident DatabaseDays–WeeksL4–L7
Vendor Status PagesReal-timeL2–L3
Layoff & Reversal TrackersDays–WeeksL6–L7
Safety DeparturesDaysL7
Regulatory & Policy FeedWeeks–MonthsL5–L7
S&P 500 AI Risk DisclosuresQuarterlyAll
Regional Salary BenchmarksAnnualL6–L7

Regional Salary Benchmark Sources

Workforce VaR uses median enterprise employee cost (salary + employer benefits) from official labour statistics and established compensation surveys in each region.

United StatesBureau of Labor Statistics (BLS) Occupational Employment Statistics
EU-27Eurostat Structure of Earnings Survey
United KingdomONS Annual Survey of Hours and Earnings (ASHE)
India (Tier 1)Mercer Total Remuneration Survey + Aon India Salary Increase Survey
JapanMHLW Basic Survey on Wage Structure
SwitzerlandFSO Labour Force Survey (LSE)
SingaporeMOM Labour Force Survey
UAEMOHRE / GulfTalent Salary Survey

What This Is Not

Not financial advice. VaR figures are indicative benchmarks for risk awareness, not forecasts or investment recommendations.

Not a real-time feed. Events are curated and updated weekly. Automated feeds are on the roadmap.

Not company-specific. Instruments are sector-level benchmarks. For your organisation's Medha Grade, book an assessment.

Not static. Grades, severity, and VaR update as new events are recorded. The framework improves with every assessment.

See it in action

The AI Risk Pulse applies this framework to live data, every week. Or get your own organisation's baseline grade before training.