CHISOKULAB| The Control Room
Methodology

How AEVA Works

Fourteen original intellectual contributions across four dimensions — developed from eleven years of field observation across nine countries and four industry verticals. Every element is implementable by a delivery team without requiring external consultancy to interpret.

DIMENSION 01

Ceremony Restructuring

Six core Agile ceremonies evolved into AI-era equivalents. Each has an original name — original AEVA intellectual property not existing in any prior methodology framework. The primary delivery unit shifts from Sprint to Increment — scope-defined, not time-defined.

PRECISION BACKLOG REFINEMENT

Replaces Backlog Refinement

  • - Core innovation: Functional-Technical AC Taxonomy — original IP
  • - Functional AC: written by PO, describes WHAT the system does, human-authored, non-negotiable
  • - Technical AC: AI-generated from Functional AC, human-reviewed by AC Architect
  • - Hallucination boundary: when Functional AC is ambiguous, AI fills the gap incorrectly. The governance solution is airtight Functional AC — not more capable AI.
  • - Pre-meeting: AI decomposes epic, generates Technical AC, scores DCI. During meeting: humans verify correctness and AC precision only. Duration: dramatically reduced.

INCREMENT PLANNING

Replaces Sprint Planning

  • - Dual-Track Capacity Model: Track 1 (DCI 4-6, AI-Executable) — bounded by QA review bandwidth, not coding velocity. Track 2 (DCI 11-16, Human) — traditional capacity calculation.
  • - New formula: Available QA review bandwidth ÷ Average AI output review time = Maximum AI-executable features per Increment
  • - Traditional: 2-4 hours estimation debate. AEVA: 45 minutes commitment and sequencing.
  • - Governance enforced upstream as infrastructure — never debated in the meeting.

PULSE SYNC

Replaces Daily Standup

  • - Layer 1 — AI Pulse (async, pre-meeting): automated dashboard — AI completion status, burn rate, blockers, QA backlog. AI generates it. Nobody presents it. Removes all status reporting from human conversation.
  • - Layer 2 — Human Sync (8 minutes synchronous): decisions required today, cross-functional collaboration, human energy signals no dashboard captures.
  • - Traditional: 15 minutes, primarily status reporting. Pulse Sync: 8 minutes, exclusively decisions and cohesion.

DATRETRO

Replaces Sprint Retrospective

  • - Phase 1: LLM pre-analyses Increment data — commit logs, velocity, blockers, defect patterns, Feature Clearance compliance rates. Generates evidence-based responses to the three retrospective questions.
  • - Phase 2: humans make decisions on evidence. "The data shows X — why did that happen, and what do we own as humans to fix it?"
  • - Explicitly includes AI output quality as a retrospective variable — no traditional retrospective does this.

INCREMENT GOVERNANCE REVIEW

Replaces Sprint Review

  • - Two-tier model: Tier 1 — continuous Human Acceptance Gate (PO reviews features as they complete, no ceremony, CI/CD principle applied to human intent validation). Tier 2 — formal 60-90 minute ceremony at Increment end.
  • - Four Tier 2 objectives: integration verification, business alignment, AI governance posture review (what did AI decide autonomously, what did humans override), DCI calibration signal.
  • - Governing principle: review cadence is human-paced, not AI-paced.

FEATURE CLEARANCE

Replaces Definition of Done

  • - Governance unit shifts: Stories are execution units. Features are governance units.
  • - Gate 1 — Technical Completion (Tech Lead + AI Output Validator): stories complete, hallucination check documented and cleared, Functional AC compliance confirmed, deployed to staging.
  • - Gate 2 — Governance Log (AI Delivery Orchestrator, max 10 minutes): DCI audit trail, AI vs human decision summary, regulated domain flag.
  • - Progressive Governance Lightening: Feature Clearance is the lightest ceremony by design — all governance work done upstream.
  • - Non-negotiable: nothing enters production without both gates cleared.

DIMENSION 02

Lean AI-Era Team Structure

Every traditional Agile role assessed against a four-question Lean Test. The goal: maximum judgment per headcount — not maximum headcount managing AI. Result: approximately 40% headcount reduction at equivalent or superior output.

Role evolution

Product Owner → Value Governance Lead

absorbs BA function with AI, owns Functional AC quality, defines AI autonomy boundaries

Scrum Master → AI Delivery Orchestrator

manages human-AI collaboration, owns Feature Clearance Gate 2

Tech Lead → AI Execution Architect

prompt engineering standards, DCI 7-10 decisions, architecture review

Senior Developer → AI Orchestrator Senior

directs AI agents, validates output at architecture level

Junior Developer → AI Orchestrator Entry

executes at dramatically higher effective velocity using AI tools

QA Engineer → AI Output Validator

hallucination detection, Functional AC compliance — Increment capacity ceiling

Business Analyst → Absorbed into PO role

Eliminated as standalone role.

Technical Architect → DCI Governance Lead

shared across teams, owns DCI calibration

Four new roles (original IP)

  • 1. AI Output Validator — does not exist in any current enterprise job architecture. Specialised QA function whose primary competency is distinguishing AI output that correctly implements human intent from output that is technically plausible but functionally incorrect. The Increment capacity ceiling for the AI Execution Track.
  • 2. AI Delivery Orchestrator — manages AI Pulse dashboard, facilitates DCI scoring, owns Feature Clearance Gate 2, primary escalation point for AI output anomalies.
  • 3. AI Governance Lead — organisation-level role (not team-level). Owns EU AI Act compliance, internal AI policy, Increment Delivery Charter. Reports to CDO or CTO.
  • 4. Acceptance Criteria Architect — ensures Functional AC is sufficiently precise to prevent AI hallucination at the execution boundary.

Junior-Senior Hierarchy Inversion

AI-assisted coding enables junior AI Orchestrators to execute at effective senior velocity on DCI 4-6 stories. Senior talent migrates to DCI 11-16 work where AI cannot substitute. Enterprises that recognise this early build higher-output teams at lower cost. This is not a future projection — it is an observable current reality.

DIMENSION 03

Velocity Recalibration and Financial Model

The primary financial frame is per-project cost — not annual team cost. AI compresses project duration. The annual comparison describes availability cost, not delivery cost.

Decision Complexity Index

The DCI replaces story points as the primary estimation framework. Story points measure human cognitive effort. The DCI measures irreplaceable human judgment — the only production variable AI cannot substitute.

Four scoring dimensions

A: Ambiguity Level

1=crystal clear → 4=fundamentally unclear, discovery required

B: Stakeholder Politics

1=single team decision → 4=multi-stakeholder negotiation

C: Domain Expertise Required

1=generic logic → 4=regulatory/compliance critical

D: Risk Consequence

1=easily reversible → 4=compliance/legal/financial consequence

Execution tiers

DCI 4-6: AI-Executable

5-8x speed multiplier realistic

DCI 7-10: AI-Assisted

2-3x speed multiplier realistic (80/20 human-AI collaboration)

DCI 11-13: Human-Led

1.2-1.5x speed multiplier realistic

DCI 14-16: Human-Only

1.2x speed multiplier realistic

Original IP — DCI Governance Shield

Three-layer architecture preventing gaming of DCI scores:

  • - Layer 1: Blind multi-scorer consensus — Tech Architect, PO, and AI Delivery Orchestrator score simultaneously without visibility of others' scores. High variance triggers mandatory public justification.
  • - Layer 2: AI independent audit score — AI scores same story independently. Variance >4 points triggers Scoring Variance Alert. Team must resolve publicly before story enters Increment.
  • - Layer 3: Historical calibration tracking — patterns of consistent score inflation across Increments trigger Scoring Pattern Inconsistency flag. Bias does not survive longitudinal delivery data.

Financial model

  • - Traditional team, 6 months: £332,500
  • - AEVA lean team, ~1 month: £38,220
  • - Per-project saving: 88% reduction
  • - Annual project capacity (same team): 2 traditional → up to 12 AEVA
  • - Annual revenue capacity at £500K fee: £1M → £6M

DIMENSION 04

Five Transition Risks

The predictable, near-universal failure patterns that destroy enterprise AI adoption programmes before they reach production scale. Not technology failures — governance, behaviour, and leadership failures. Named, defined, and mitigated in AEVA.

Risk — GOVERNANCE LAG

Definition: The systematic selective interpretation of AEVA governance elements — retaining efficiency-generating components while discarding governance-intensive ones. Creates a facade of AI delivery maturity without the substance.

Early warning: "That gate is optional for our context." Feature Clearance compliance falling below 100%.

Mitigation: Feature Clearance gates structurally tied to Increment continuation — not advisory. Gate removal requires documented decision reviewed by AI Governance Lead.

Risk — GOVERNANCE DRAG

Definition: The paradox where governance infrastructure designed to enable safe AI delivery becomes the primary bottleneck — eliminating AI efficiency gains entirely.

Key insight: AI can generate governance documentation at machine speed. Requiring more documentation creates overhead that humans review at human speed. The bottleneck worsens, not improves.

Mitigation: Governance time ceilings as misconfiguration signals. Feature Clearance Gate 2: 10 minutes maximum. Proportionality Rule: governance intensity proportional to DCI risk score.

Risk — SHADOW AI PROLIFERATION

Definition: The parallel ungoverned AI delivery stream where employees use unsanctioned AI tools on uncleared data — undermining every governance mechanism from the inside.

Directly observed: Yamaha Motor Solutions India. Employees using ChatGPT via personal devices on uncleared data with no governance. Leadership projecting purposeful AI adoption externally. The gap was structural.

Mitigation: Process-Level Immunity — precise Functional AC as a universal quality gate. Any output must comply with Functional AC to pass Feature Clearance, regardless of which tool generated it. Governance is output-focused, not tool-focused.

Risk — THE IDENTITY CRISIS

Definition: Systematic quiet resistance to AI adoption by experienced practitioners whose professional identity is built on skills AI has disrupted. Manifests not as open objection but as invisible obstruction indistinguishable from diligence.

Pattern: Technical Architect consistently scores every story DCI 14-16. Senior developer insists on personally reviewing all AI output creating a single-point bottleneck. Deniable. Rational. Quietly catastrophic.

Mitigation: New Crown Reframe: Old identity — "I build things others cannot" (threatened by AI). New identity — "I govern what AI builds. I make decisions AI cannot make. I am the reason AI output is safe, intentional, and auditable" (made more valuable by AI).

Risk — STRATEGIC BLINDNESS

Definition: Structural misalignment between technology and governance leadership — producing AI adoption decisions that are technically plausible but strategically incoherent, creating liability at scale while appearing to create value.

Root cause: tech leadership optimises for delivery velocity. Governance leadership optimises for risk reduction. Structurally opposed. AI widens the gap to breaking point.

Strategic Adoption Principle

AI capability without governance infrastructure is not digital transformation. It is liability creation at scale. Investment sequence: (1) governance framework, (2) team training, (3) AI tool deployment, (4) scale. Never inverted.

Intellectual Property

Fourteen original IP claims across ceremony architecture, governance systems, team structure, financial modelling, and risk taxonomy. Filed under provisional patent protection.

📜

Provisional Patent Pending

Filed with the Indian Patent Office, 2026. Fourteen claims covering ceremony architecture, DCI framework, DCI Governance Shield, team structure, financial model, and Five Transition Risk taxonomy.

📄

Copyright Registration

Copyright Diary No. LD-14934/2026-CO. Registered under Agam Agrawwal, 2026. Reproduction prohibited without written consent.

✈️

Production Deployment

AEVA is not theoretical. It has been deployed in production enterprise environments across manufacturing (Yamaha Motor Solutions India) and aviation (Etihad Airways, Dubai Airports).

See how AEVA applies to your delivery environment

View Deployment Evidence →Book a Call →