Pillar 2: Data & AI Readiness

Overview

Pillar 2 answers the question: Is the data ready for these specific agents?

This pillar translates prioritized use cases from Pillar 1 into concrete data requirements and readiness assessments—without triggering enterprise-wide data transformation.

---

Why Pillar 2 Matters

The Problem We're Solving

Most organizations approaching AI data readiness fall into one of two traps:

1. Enterprise Transformation Paralysis — "We need to modernize our data platform before we can do AI." Years of infrastructure work before any value delivery.

2. Naive Assumptions — "Our data is fine." Hidden quality issues, access constraints, and governance gaps surface during implementation, causing delays and rework.

Pillar 2 breaks both patterns by providing agent-specific data assessment that surfaces risk early without requiring transformation.

What Success Looks Like

By the end of Pillar 2, the client organization has:

• Data readiness assessments for each prioritized agent
• Clear data ownership with named human owners
• Explicit understanding of what agents can read vs. propose to change
• Documented data risks and validation requirements
• Entry criteria met for Pillar 3 (Governance)

---

The OAIO Methodology: Agent-Specific Data Readiness

Pillar 2 deliberately avoids enterprise-wide data assessment. Instead, Orion runs a series of agent-specific working sessions, one agent at a time, each with its own focused team.

Step 1: Session Structure and Sequencing

See Data Readiness Session Guide

Each agent gets its own dedicated working session (4-5 hours). Sessions are run on sequential days to maintain focus.

The Critical Rule: The persona leader responsible for Agent Case 1 is NOT in the room for Agent Case 2. This prevents:

• Scope bleed between agents
• Political compromise on requirements
• Cross-contamination of priorities

Required Personas (Per Session):

Persona	Role	Why They're Essential
Agent Business Owner	LOB leader accountable for the problem	Owns prioritization, tradeoffs, adoption success
Veteran Practitioners (2-3)	Domain experts	Understand how data is actually created, corrected, interpreted
System Owners	IT application owners	Explain how data is stored and accessed in practice
Data Stewards	DBA, BI, Analytics leads	Validate feasibility, clarify data structure and lineage
OAIO Facilitators	Orion data & agent experts	Challenge assumptions, translate to agent requirements

Explicitly Excluded: Other agent owners, security, legal, finance, enterprise architecture, platform modernization teams.

---

Step 2: Data Sketching Exercise

The core method of each session is a literal sketching exercise. Orion facilitates whiteboard or digital mapping to visualize:

What We Map:

• Upstream and downstream data sources
• Internal vs. external data sources (clearly separated)
• Human touchpoints where data is created or corrected
• Decision points that rely on data interpretation

Internal vs. External Data:

Category	Examples	Key Considerations
Internal	Systems of record, operational databases, document stores	Access control, data quality, ownership
External – Public	Regulations, standards, public records	Update frequency, reliability, licensing
External – Private	Partners, clients, licensed third-party data	Contracts, usage rights, refresh mechanisms

---

Step 3: Data Tagging

For every identified data source, the team explicitly tags:

Tag	Question	Why It Matters
Location	Where does this data live?	System of record, file store, tool
Access Method	How is it retrieved?	API, batch, manual export, read-only view
Sensitivity	How sensitive is it?	Public, internal, confidential, regulated
Existing Controls	What controls exist?	Permissions, approvals, audit logs
Agent Interaction	Read or write?	Critical for governance in Pillar 3

Agent Interaction Model (Most Critical Tag):

For each data source:

• Does the agent only read this data?
• Or does the agent propose changes?
• If changes are proposed, who approves them?

---

Step 4: Human Ownership Assignment

Every data source is paired with a named human owner responsible for:

• Access decisions
• Quality accountability
• Audit responsibility

The Named Data Access Owners Register documents:

• Data source name
• Human owner (by name, not role)
• What access they can grant
• What approval process exists
• Contact information

---

Session Agenda

See Data Readiness Session Guide for detailed facilitation guidance.

Per-Agent Session (4-5 Hours):

Time	Focus	Purpose
0:00–0:30	Agent Context & Success Definition	Reconfirm problem statement, define adoption success
0:30–2:00	Data Source Identification & Sketching	Map all data sources, identify human touchpoints
2:00–2:15	Break
2:15–3:30	Data Tagging & Risk Surfacing	Apply tags, discuss agent read vs. write behavior
3:30–4:30	Feasibility & Scope Adjustment	Identify unacceptable risks, adjust scope
4:30–5:00	Session Wrap-Up	Summarize assumptions, confirm validation needs

---

Outputs and Handoffs

Pillar 2 Deliverables (Per Agent)

Deliverable	Description	Example
Agent Data Map	Visual representation of data dependencies	View Example
Tagged Data Inventory	Location, access, sensitivity, controls for each source	View Example
Named Data Access Owners Register	Human owners for each data source	View Example
Agent Access Model	Read vs. propose-change permissions	View Example
Open Risk & Validation Log	Assumptions to be tested, risks identified	View Example

Handoff to Pillar 3

Pillar 2 outputs flow directly into Pillar 3 (AI Protection & Operational Trust):

• Data sensitivity mapping informs governance requirements
• Agent access models shape permission frameworks
• Named owners become accountability touchpoints
• Risk logs inform escalation procedures

---

The Data Readiness Scoring Framework

Each data source is scored across four dimensions:

Dimension	0 (Not Ready)	1 (Major Work)	2 (Minor Work)	3 (Ready)
Accessibility	No access path	Access requires major work	Access path exists, minor improvements needed	API or reliable access available
Quality	Data unreliable	Significant quality issues	Minor quality gaps	Data meets agent requirements
Governance	No controls	Controls inadequate	Controls exist, enhancements needed	Governance appropriate for use case
Integration	No integration path	Major integration effort	Standard integration work	Ready to connect

Composite Score Interpretation:

Score	Status	Implication
10-12	Ready	Proceed to Pillar 3
7-9	Minor remediation	Short-term fixes before proceeding
4-6	Significant work	Consider prioritization adjustment
0-3	Not feasible	Data initiative required or deprioritize

---

Common Patterns

Pattern: Shadow Data

Organizations often discover critical data exists outside official systems—in spreadsheets, personal databases, or undocumented sources. Pillar 2 surfaces this shadow data and determines whether to:

• Formalize it into official systems
• Accept it with documented risk
• Adjust agent scope to avoid dependency

Pattern: Quality Debt

Historical data quality issues compound over time. Pillar 2 helps organizations decide:

• Remediate existing data (expensive, time-consuming)
• Establish quality gates for new data (faster, but historical data unchanged)
• Scope agents to work with current quality (accept limitations)

Pattern: Governance Gaps

Data may be technically accessible but legally or ethically constrained. Pillar 2 identifies these constraints early, before they block implementation.

Pattern: External Data Dependencies

Agents often need external data (regulations, market data, partner feeds). These dependencies introduce:

• Refresh timing issues
• Licensing constraints
• Reliability concerns

---

Relationship to Pillar 1

Pillar 2 may cause you to revisit Pillar 1 priorities. This feedback loop is intentional.

If a high-priority use case has severe data readiness gaps:

1. Invest in data remediation — If the value justifies the effort
2. Deprioritize — In favor of more data-ready alternatives
3. Phase the approach — Start with available data, expand scope later

---

Delivery Timeline

---

Common Pitfalls

---

Facilitator Guidance

Preparation

1. Review Pillar 1 outputs — Understand each agent's business rationale and success criteria
2. Pre-identify data sources — Request preliminary data inventories before sessions
3. Confirm participants — Ensure veteran practitioners and system owners are committed
4. Prepare sketching materials — Whiteboard, digital tools, or templates

Delivery Tips

Opening:

• Reframe immediately: "This is about whether this agent can work with data as it exists today—not about changing your data architecture."
• Establish the sketch as the central artifact—everyone contributes

Managing the Room:

• Prevent system owners from over-explaining architecture—stay focused on what the agent needs
• Draw out veteran practitioners—they know the real data flows
• Challenge "our data is fine" statements with specific scenarios

Closing:

• Summarize the sketch and tags visually
• Confirm named owners for each data source
• Be explicit about risks and next steps

---

Pricing and Positioning

Scope Options

Scope	Duration	Description
Rapid assessment	1-2 weeks	High-level readiness check for top 3 agents
Comprehensive assessment	3-4 weeks	Detailed analysis for 5+ agents with remediation roadmap
Embedded assessment	Ongoing	Data readiness integrated with implementation

Cloud Partner Integration

Pillar 2 assessments often surface cloud-related data requirements:

• AWS — S3 access patterns, Glue catalog integration, Bedrock knowledge base compatibility
• Microsoft — Azure data services, Fabric integration, Copilot Studio data requirements

---

Required Collateral

---

Reference Materials

Facilitator Guides

• Data Readiness Session Guide — Full session design with agenda, personas, and outputs

Related Content

• NorthRidge Case Study: Pillar 2 — Story-based walkthrough
• Pillar 1: Value & Adoption — Prerequisites for Pillar 2
• Pillar 3: AI Protection & Operational Trust — Where Pillar 2 outputs flow

External Resources

Data Governance:

• DAMA-DMBOK — Data Management Body of Knowledge
• Data Governance Institute — DGI Data Governance Framework

Cloud Data Services:

• AWS Data Lake Architecture — Reference architecture for enterprise data lakes
• Azure Data Governance — Microsoft data governance solutions
• Google Cloud Data Governance — Data governance on Google Cloud

Privacy & Compliance:

• GDPR Official Text — EU General Data Protection Regulation
• CCPA Official Text — California Consumer Privacy Act