AskAjay.ai | Framework Worksheet
Signature Framework

PRIME Framework
Implementation Worksheet

A structured 90-day implementation tool for wiring responsible AI directly into your development pipeline. Covers all five PRIME dimensions, stack mapping, KPI baselines, and sprint planner.

By Ajay Pundhir askajay.ai/tools Version 1.1 · Updated 2026-05-02

Canonical reference: The Real Cost of AI Isn’t Ethics — It’s Rework

How to Use This Worksheet

  1. Pick one revenue-adjacent generative AI use case — the one your CEO mentions in earnings calls. Run PRIME against it before expanding to others.
  2. Work through the five PRIME dimensions (Purposeful Impact → Respect for Humans → Interpretable by Design → Mitigated by Engineering → Enforceable Accountability) for your chosen use case.
  3. Map PRIME to your stack across five operational surfaces: Governance, Data, Model & Agent, Experience, Operations.
  4. Set KPI baselines for Intelligence Velocity (time-to-resolve high-severity AI issues) and Principles-to-Product Coverage (P2P) — the percentage of shipped AI features with completed PRIME artifacts. These are the two metrics that convert responsibility into engineering discipline.
  5. Use the 90-day sprint planner to assign owners and track progress through Baseline → Instrument → Scale.
PRIME means foundational — the thing you do first, not the thing you bolt on after.
The upfront investment in PRIME artifacts is a fraction of the rework cost you're already paying. This worksheet turns the OECD's values-based principles into engineering requirements that live where the code lives.
Brand commitments behind this worksheet

Responsible AI is foundational, not bolt-on. PRIME artifacts go into sprint one, not sprint twelve. A fairness evaluation is a test suite, not a six-month study. A benefit hypothesis is a paragraph, not a dissertation. The rework you avoid more than pays for the artifacts you produce.

Two KPIs, not twenty. Intelligence Velocity and Principles-to-Product Coverage are sufficient to tell your board whether trust is improving or eroding. If your responsible AI dashboard has more than five KPIs, you’re measuring effort, not outcomes. Resist metric proliferation.

Section 1: Use Case Selection

Select one revenue-adjacent generative AI use case as your PRIME pilot. Focus beats breadth.

Field Your Response
AI Use Case Name
Business Owner
Revenue Adjacency
Why this use case matters commercially
Deployment Stage
Pilot / Production / Scaling
Users / Audience
Who consumes the AI's outputs?
Current Responsible AI Coverage
What exists today? Policy doc, review board, nothing?

Section 2: PRIME Dimension Assessment

For your selected use case, assess each PRIME dimension. Check what exists, identify gaps, define the artifact to produce, and name the owner.

P
Purposeful Impact — Start with who benefits
OECD Principle: Inclusive Growth, Sustainable Development and Well-Being

Current State Checklist

Benefit hypothesis defined and attached to the user story / ticket
Affected-groups map completed (who benefits, who might be disadvantaged)
Measurable outcomes defined (cycle time, accessibility gains, error reduction)
Outcomes included in definition of done

Gap Identification

Artifact to Produce

ArtifactOwner (Name)Target Date
Benefit Hypothesis
Affected-Groups Map
R
Respect for Humans — Codify where humans stay in the loop
OECD Principle: Human-Centred Values and Fairness

Current State Checklist

High-stakes contexts pre-registered (employment, health, credit, protected categories)
Decision rights documented explicitly
Opt-out and correction channels built into UX
Plain-language disclosures designed from sprint one

Gap Identification

Artifact to Produce

ArtifactOwner (Name)Target Date
High-Stakes Context Register
Decision Rights Document
Opt-Out / Correction UX Spec
I
Interpretable by Design — Ship explanations as features
OECD Principle: Transparency and Explainability

Current State Checklist

Model cards maintained as living documents
Data sheets and prompt/version lineage tracked
Source citations and confidence levels shown in UI
Explanation surfaces calibrated per audience (clinician / consumer / regulator)

Gap Identification

Artifact to Produce

ArtifactOwner (Name)Target Date
Model Card (living document)
Data Sheet / Lineage Record
Explanation UX per Audience
M
Mitigated by Engineering — Treat safety like SRE
OECD Principle: Robustness, Security and Safety

Current State Checklist

Automated evaluations for hallucination, toxicity, prompt injection, jailbreak resilience
Release gates with hard thresholds (e.g., jailbreak success rate ≤ 1%)
Kill switches bound to live metrics, not manual intervention
Privacy leakage and license violation checks running in CI/CD

Gap Identification

Artifact to Produce

ArtifactOwner (Name)Target Date
Evaluation Harness (halluc., toxicity, jailbreak)
Release Gate Thresholds
Kill Switch Configuration
CI/CD Privacy & License Checks
E
Enforceable Accountability — Name the owner, keep the evidence
OECD Principle: Accountability

Current State Checklist

Accountable owner assigned (a name, not a team) for every AI feature
Evidence trail defined upfront: datasets, prompts, model versions, eval results, production logs
Policies encoded as code (audits are reproducible, not archaeological)
Vendor contracts include evidence artifacts and incident-sharing clauses

Gap Identification

Artifact to Produce

ArtifactOwner (Name)Target Date
Accountability Register (owner per feature)
Evidence Trail Specification
Policy-as-Code Configuration
Vendor Contract Clauses
PRIME Dimension Checkpoint You should now have a current-state assessment and gap analysis for all five PRIME dimensions, plus a list of artifacts to produce with named owners. If any dimension has zero checkboxes ticked, that's your highest-priority gap.

Section 3: PRIME Across the Stack

Map your PRIME implementation to the five operational surfaces. For each surface, document what exists today and what needs to be built.

Operational Surface Current State Target State What Gets Measured
Governance
Policy-as-code, provenance tags
P2P Coverage per PRIME dimension
Data
Contracts, provenance, consent
Data lineage completeness; leakage incident rate
Model & Agent
Approved models, eval harness
Intelligence Velocity; adversarial test pass rate
Experience
Disclosures, feedback, explanation UX
Explanation coverage; user correction rate
Operations
Incident registry, weekly triage
Incidents detected; mean time to mitigation

Section 4: KPI Baselines

The two metrics that convert responsibility into engineering discipline. Measure your starting position, then track monthly.

Intelligence Velocity
Elapsed time from risk discovery to fleet-wide mitigation
Your current IV (hours/days)
Target: < 72 hours (high-severity)
P2P Coverage
% of shipped AI features with completed PRIME artifacts
Your current P2P (%)
Target: > 80% by end of Q1

Monthly KPI Tracker

Month Intelligence Velocity P2P Coverage (%) Notes / Actions
Baseline (Day 0)
Month 1
Month 2
Month 3
Month 6
KPI Checkpoint If your responsible AI metrics dashboard has more than five KPIs, you're measuring effort, not outcomes. Intelligence Velocity and P2P Coverage are enough to tell your board whether trust is improving or eroding.

Section 5: 90-Day PRIME Sprint Planner

Assign ownership, dates, and accountability for each month.

Month 1
Baseline: Adopt & Map Adopt OECD Principles as your standard. Map them to the NIST GenAI Profile for a control catalog. Build the minimum viable control book for your chosen use case. Wire evidence checks into CI pipeline (warn, not block). Complete PRIME dimension assessment above.
Owner:
Target completion date:
Month 2
Instrument: Build & Measure Build evaluation harness (hallucination, toxicity, jailbreak). Stand up incident registry (OECD AIM taxonomy). Start weekly triage (engineering + product + risk, 30 min). Switch pipeline gates from warn to block for high-severity. Measure first Intelligence Velocity number.
Owner:
Target completion date:
Month 3
Measure & Validate Run the full KPI cycle for the pilot. Capture Intelligence Velocity (time-to-resolve high-severity AI issues) and P2P Coverage (% of shipped features with all five PRIME artifacts). Draft a one-page EU AI Act readiness note — risk category hypothesis plus the evidence you already collect. Review vendor contracts for PRIME clauses. The numbers you collect here drive the Day 90 scale decision below.
Owner:
Target completion date:
Day 90
Checkpoint: Scale Decision Review KPIs. If IV < 72 hours and P2P > 80%, expand PRIME to second use case. If not, iterate on the pilot. Document lessons learned.
90-day review date:
Scale decision & next priorities:

Section 6: PRIME vs. Existing Tools Audit

Which responsible AI tools are already in use? Where does PRIME fill the gaps?

Tool / Framework In Use? What It Covers PRIME Gap It Doesn't Fill
Microsoft HAX Toolkit UX design patterns, 18 Guidelines for Human-AI Interaction No CI/CD integration, no automated evals, no policy-as-code
Google SAIF AI security, threat taxonomy, adversarial robustness No fairness, explainability, benefit assessment, or accountability
NIST AI RMF Comprehensive risk management framework (Govern, Map, Measure, Manage) PRIME provides the pipeline-level implementation NIST describes at policy level
Internal Tools
Other
PRIME doesn't replace these frameworks — it connects them.
HAX informs your UX decisions. SAIF informs your threat assessment. NIST provides the control catalog. PRIME is the connective tissue that turns design intentions and security assessments into measurable, enforceable, sprint-level engineering practice.

PRIME Readiness Scorecard

Rate your completion of each PRIME dimension. Circle: Not Started / In Progress / Complete.

P — Purposeful Impact
Not Started
In Progress
Complete
R — Respect for Humans
Not Started
In Progress
Complete
I — Interpretable by Design
Not Started
In Progress
Complete
M — Mitigated by Engineering
Not Started
In Progress
Complete
E — Enforceable Accountability
Not Started
In Progress
Complete

Glossary

Definitions used throughout this worksheet. These align with the canonical PRIME article.

PRIME. A five-dimension framework for wiring responsible AI directly into the development pipeline: Purposeful Impact, Respect for Humans, Interpretable by Design, Mitigated by Engineering, Enforceable Accountability. Maps OECD AI Principles to NIST AI RMF controls.

Intelligence Velocity (IV). The time it takes to resolve a high-severity AI issue (data drift, fairness regression, prompt injection, hallucination) once detected. Target: under 72 hours for the pilot use case. Lower is better.

Principles-to-Product Coverage (P2P). The percentage of shipped AI features with all five PRIME artifacts complete: benefit hypothesis, fairness evaluation, explanation UX, safety gates, and audit trail. Most organizations start at 20–25%. Quarter-one target: 80%.

PRIME pilot use case. The single revenue-adjacent generative AI use case selected for first PRIME implementation. Focus beats breadth — one use case run end-to-end is more valuable than five run partially.

Stack mapping. Translating PRIME requirements into concrete artifacts at five operational surfaces: Governance, Data, Model & Agent, Experience, Operations. Frameworks that stay conceptual don’t survive contact with engineering teams.

Rework cost. The percentage of each engineering sprint consumed by remediating responsible-AI gaps caught after deployment. Pre-PRIME baseline observed across engagements: ~30% of sprint capacity. Post-PRIME at month 3: under 5%.

EU AI Act readiness note. A one-page document produced at Month 3 stating the pilot’s risk-category hypothesis under the EU AI Act, plus the evidence already collected via PRIME artifacts. Most of the regulatory evidence requirement is met by the artifacts you already produce.

Evidence Base

This worksheet is the operational layer of a published, sourced framework. The full argument, OECD/NIST mapping, and the financial-services case study live in the canonical article below.

Canonical article: The Real Cost of AI Isn’t Ethics — It’s Rework — the rework problem, the PRIME framework, OECD-to-NIST mapping, two-KPI rationale, and a 90-day case study (rework cut from 30% to under 5%).

Companion frameworks:

Foundational external sources: OECD AI Principles · NIST AI Risk Management Framework · NIST AI 600-1 (Generative AI Profile)

Notes & Observations