Bonus 3

Templates & Checklists

Ten one-page worksheets you can copy into a doc, a wiki, or a whiteboard session. They turn the ideas in the book into artifacts your team can actually share.

Template · four-c-canvas

4 C's Architecture Canvas

A one-page canvas to evaluate or design any AI initiative against Context, Control, Cost, and Choice.

Context
  • Top business terms this agent depends on
  • Where each term is defined (system, owner)
  • Gaps in meaning the team currently fills informally
Control
  • Data this agent can read
  • Actions this agent can take
  • Approval thresholds and reviewers
Cost
  • Target unit cost per interaction
  • Expected daily / weekly volume
  • Routing and caching plan
Choice
  • Vendors and tools used
  • Abstractions that protect from lock-in
  • Exit cost estimate per vendor
Template · context-layer-design

Context Layer Design Template

Use this to scope your first context layer build.

Domain
  • Chosen business domain
  • Why this domain first
  • Boundary of scope
Metrics
  • Top 10 metrics
  • Owner per metric
  • Source table per metric
Glossary
  • Top 20 terms
  • Owner per term
  • Edge cases / exceptions
Consumer
  • First consumer (agent or dashboard)
  • Go-live date
  • Success criterion
Template · agent-governance-checklist

Agent Governance Checklist

Use before any agent goes to production.

Access
  • Identity propagation works end to end
  • Row / column policies tested
  • Sensitive data tagged
Action
  • Inventory of agent actions
  • Risk class per action
  • Policy-as-code in place
Audit
  • Structured trail for every action
  • Retention meets regulation
  • Compliance can self-serve queries
Template · ai-cost-worksheet

AI Cost Control Worksheet

A worksheet for setting and defending unit cost targets.

Cost drivers
  • Tokens in / out
  • Tool call frequency
  • Retry rate
  • Model tier mix
Targets
  • Cost per interaction target
  • Maximum tail cost
  • Monthly spend cap
Levers
  • Routing rules
  • Caching strategy
  • Prompt size budget
  • Retrieval depth budget
Alerts
  • Cost regression alert
  • Volume anomaly alert
  • Per-use-case dashboard
Template · model-choice-tree

Model Choice Decision Tree

Walk every workload through this tree before picking a model.

Step 1 — Classify the task
  • Simple classification?
  • Standard QA?
  • Complex reasoning?
  • Specialized domain?
Step 2 — Set the constraints
  • Latency target
  • Quality target
  • Cost target
  • Pick the two hard constraints
Step 3 — Choose the model tier
  • Tiny model for simple
  • Medium for standard
  • Large for complex
  • Specialized / fine-tuned for domain
Step 4 — Plan the fallback
  • When does it escalate?
  • What is the fallback model?
  • Who is notified on escalation?
Template · ai-readiness-checklist

Enterprise AI Readiness Checklist

Score every dimension from 1 (weak) to 5 (strong). Readiness is the lowest score.

The four C's
  • Context — definitions, ontology, lineage
  • Control — access and action policies
  • Cost — measurement and targets
  • Choice — portability and exit cost
Operations
  • Ownership
  • Monitoring
  • Rollback
Template · human-review-ladder

Human Review Ladder

Place every agent action on the right rung. Revisit quarterly.

Rungs
  • Rung 1 — Full automation, audit later
  • Rung 2 — Human notified
  • Rung 3 — One-click approval
  • Rung 4 — Human review and edit
  • Rung 5 — Human owns the decision
For each action
  • Action name
  • Current rung
  • Evidence to move down a rung
Template · agent-permission-map

Agent Permission Map

A grid of agents x actions x conditions. The source of truth for what can happen.

Rows
  • List of agents in production
Columns
  • Read actions
  • Write actions
  • Communication actions
  • Financial actions
Cells
  • Allowed / Denied / Approval-required
  • Reviewer if approval-required
  • Last review date
Template · platform-mapping

Platform Mapping Worksheet

Map every concept to every platform you might touch. A career-long asset.

Concepts
  • Storage
  • Table format
  • Catalog
  • Query engine
  • Semantic layer
  • Model serving
  • Governance
  • Lineage
  • Observability
Platforms
  • Databricks
  • Snowflake
  • AWS
  • Azure
  • GCP
  • Open source
Template · team-discussion-guide

Team Discussion Guide

Run this in a 60-minute team meeting after reading the book.

Opening (10 min)
  • Each person names the chapter that hit hardest
  • Each person names one habit they want to change
The four C's check (20 min)
  • Score the team's strongest project on each C
  • Identify the weakest C across the portfolio
Commitments (20 min)
  • One concrete change to start this month
  • One concrete change to stop this month
  • Owner and date for each
Close (10 min)
  • Pick the next chapter to discuss
  • Schedule the next session