AI for the Modern World

Module 7 of 10

Module 07: AI Strategy for Non-Technical Leaders

8 min read1,552 words
What you'll learn
Draft a one-page AI strategy thesis tied to measurable business or mission outcomes.Build a business case with costs, benefits, risks, and alternatives (including "do less").Sequence a 12-month roadmap with pilots, governance, and scaling criteria.Lead change management conversations that reduce fear and shadow IT.Align IT, legal, communications, and frontline staff around shared vocabulary.Run an executive workshop agenda without technical slides.

"Strategy without tactics is the slowest route to victory. Tactics without strategy is the noise before defeat." — Sun Tzu (paraphrased often—your AI plan needs both)

Opening scenario: The board asks for an "AI strategy" slide. Vendors flood your inbox. Employees already use shadow tools. IT worries about data leaks. You need a roadmap: business case, prioritized use cases, budget framing, and change management—without pretending you will "build models" in-house. This module is for managers, entrepreneurs, and professionals who must steer.

Learning Objectives

By the end of this module, you will be able to:

  • Draft a one-page AI strategy thesis tied to measurable business or mission outcomes.
  • Build a business case with costs, benefits, risks, and alternatives (including "do less").
  • Sequence a 12-month roadmap with pilots, governance, and scaling criteria.
  • Lead change management conversations that reduce fear and shadow IT.
  • Align IT, legal, communications, and frontline staff around shared vocabulary.
  • Run an executive workshop agenda without technical slides.

Concept:
An AI strategy is not a model roadmap — it is which outcomes you pursue, which risks you refuse, and which pilots earn the next dollar of spend.

1. Strategy First: Outcomes, Not Algorithms

Thesis pattern: "We will use AI to [move metric X / reduce risk Y / improve experience Z] by [assistive workflows] while [non-negotiable safeguards] remain intact."

Bad thesis: "We will adopt AI."
Better thesis: "We will reduce proposal turnaround from 10 days to 6 by assistive drafting and templated reviews, without automating legal sign-off."

Real World Example

A mid-size nonprofit writes a one-sentence thesis: “Cut grant-reporting time 30% with assistive drafting; no automated eligibility decisions.” Vendors stop waving “AI” and start showing workflow fit. Clarity saves months.

Real-world examples

  • Regional bank (illustrative): Focuses on internal efficiency first; customer-facing automation year two after governance.
  • K–12 district: Prioritizes teacher support and operations; student-facing tools only with curriculum alignment.
  • Law firm: Allows marketing and knowledge management pilots; client work under partner protocols.
LensWhat It DoesTry It Here
ThesisStates outcome + safeguardWrite 3 variants; pick one
Business caseCosts, benefits, rangesBack-of-envelope low/mid/high
RoadmapPilots before platformsOne-page Q1–Q4 with kill triggers

Fun Fact
Boards often want an “AI slide” — the winning leaders bring metrics and red lines, not logos.

Try This Now

  • Write three thesis variants; pick one; ask ChatGPT to stress-test assumptions—you decide what's fair.

Table: Strategy anchors

AnchorQuestion
CustomerWho benefits externally?
EmployeeWho burns out if we skip this?
RiskWhat cannot break?
DifferentiationAre we copying hype or competing?

Discussion prompts

  • What will you NOT automate this year—and why?

2. Business Case: Numbers, Narrative, and Honesty

Components

  • Problem (pain quantified if possible)
  • Proposed approach (tools, workflows, owners)
  • Costs (licenses, training, integration, opportunity cost)
  • Benefits (time, revenue, quality, compliance)
  • Risks (privacy, errors, vendor lock-in)
  • Alternatives (hire, outsource, wait)

Simple ROI: (Annualized benefit − annualized cost) / cost. Pair with qualitative benefits (speed to market, morale).

Scenario: A nonprofit imagines huge savings; forgets training. Reality: Year-one net positive small; Year-two better if adoption sticks.

Try This Now

  • Build a back-of-envelope table with ranges (low/mid/high) for one pilot.

Table: Cost categories people forget

Hidden costExample
Change managementWorkshops, FAQs
MonitoringSomeone reads logs
IntegrationSSO, MDM
ExitExport pain

Discussion prompts

  • Who signs the business case if benefits are uncertain?

3. Roadmap: Pilots, Scale, and Kill Switches

Phase 0 — Discovery (4–8 weeks): inventory shadow use, data classes, quick wins.
Phase 1 — Pilots (8–12 weeks): one internal, one external-facing (if appropriate); metrics.
Phase 2 — Scale: playbooks, training, support.
Phase 3 — Optimize: cost reviews, vendor competition, advanced workflows.

Kill switch triggers: spike in errors, privacy incident, regulatory change, adoption < threshold.

Walkthrough: roadmap workshop (90 minutes)

  1. Stakeholder map (15 min)
  2. Pain/ gain sticky notes (25 min)
  3. Pick two pilots (20 min)
  4. Risks & owners (20 min)
  5. Next 30 days (10 min)

Try This Now

  • Draft milestones for Q1–Q4 on one page; share with a critic.

Discussion prompts

  • What evidence would make you pause scaling?

4. Change Management: Fear, Skill, and Fairness

Fears: job loss, surveillance, unfairness, looking obsolete.
Responses: transparency, training, re-skilling budget, clear roles ("AI assists; you decide").

Practices

  • Champions in each department
  • Office hours
  • Safe sandboxes
  • Recognition for good prompt hygiene

Real-world examples

  • Hospital admin: Nurses consulted before note tools; workflow adjusted to reduce clicks.
  • Agency: Junior staff feared replacement; leadership reframed as editorial uplift with training.

Try This Now

  • "Draft a compassionate FAQ for staff worried about AI—5 Qs." Edit heavily to match your tone.

Table: Communication channels

ChannelBest for
All-handsPrinciples
Team meetingsSpecific workflows
Written policyRed lines

Discussion prompts

  • How will you measure trust—not just usage?

5. Governance Lite: Enough to Move Safely

Minimum viable governance

  • Data classification cheat sheet
  • Approved tools list + request process
  • Incident reporting path
  • Retention defaults
  • Review calendar (quarterly)

Roles: Executive sponsor, IT/security, legal/privacy, function owners.

Try This Now

  • Create a RACI (Responsible, Accountable, Consulted, Informed) for one pilot.

Discussion prompts

  • When does governance become bureaucracy—and how will you tell?

Activities

  1. Strategy tweet: One sentence strategy; peer critique.
  2. Business case swap: Review another learner's ROI assumptions.
  3. Roadmap draft: 4 quarters, two pilots, metrics.
  4. Town hall outline: 15-minute non-technical talk.
  5. Risk register: 10 risks with mitigations and owners.

Your Challenge

Produce an AI Strategy Mini-Brief (3 pages max):

  • Thesis + non-goals
  • Two pilots with metrics and budget bands
  • 90-day plan + governance skeleton
  • Stakeholder comms plan (who hears what, when)

Present to a mentor or manager; capture three edits.

Key Takeaways

Try This!
Facilitate a 10-minutewill not automate” round with your team — capture five non-goals for the year. Post them beside the pilot list.

  • Strategy ties AI to outcomes and safeguards—not logos.
  • Business cases need honest costs and ranges, not precision theater.
  • Roadmaps sequence pilots before platform promises.
  • Change management is the multiplier—tools are cheap, trust is not.
  • Governance can be light but explicit.
  • Kill switches are a sign of maturity, not failure.

Resources

  • NIST AI Risk Management Framework — overview pages for leaders.
  • ISO/IEC 42001 — management system standard (orientation with legal/compliance).
  • Internal IT architecture contact.
  • Module 08 ethics and Module 09 security for depth.

Executive One-Pager Template (Outline)

  • Why now (market, risk of standing still)
  • Where we play (internal assist first, etc.)
  • How we win (pilots, metrics)
  • Risks & mitigations (top 3)
  • Decisions needed (budget, tools, policy)
  • Next 30 days (named owners)

Comparison Table: Build vs. Buy vs. Partner

PathWhen it fits
Buy SaaSStandard workflows
PartnerNeed customization
Build (rare for this audience)Core differentiator + capital

Scenario: Board Pushback

Board: "Competitors announced AI." You: "Here's our pilot metric and risk controls; we launch internal assist in 30 days rather than rushing customer claims."

Discussion Prompts (Seminar)

  • What skills will you hire for in two years if drafting is commoditized?
  • How do you avoid AI theater (demos without adoption)?
  • Which vendor lock-in risks are acceptable for your size?

Reflection Journal

  1. My top outcome for AI this year: ___
  2. My red line: ___
  3. The stakeholder I most need to align: ___

Facilitator Agenda: Board AI Briefing (25 Minutes)

  • 0–3: Definition (Module 01 one-liner)
  • 3–10: Two pilots
  • 10–18: Risks/governance
  • 18–25: Decisions & timeline

Anti-Patterns to Name Aloud

  • Pilot theater without metrics
  • Shadow IT ban without approved alternatives
  • Vendor roadmap confused with your strategy

Closing

Leadership in the AI era is clarity: fewer buzzwords, more accountability, sequencing, and care for people carrying the change.

Bonus: "Strategy Stress Test" Prompts (Use with Any Assistant)

Ask: "List 10 ways this AI strategy could fail in the first 90 days; include organizational, technical, and vendor causes."
Then ask: "For the top three failures, propose an early warning indicator we could track weekly."
Human: delete generic fluff; keep signals you can actually observe (help desk tickets, edit rates, survey items).

Bonus Table: Pilot Selection Scorecard

CriterionWeight (example)Notes
ImpactHighTie to KPI
FeasibilityHighData ready?
RiskMediumSensitive?
Learning valueMediumTeaches org
VisibilityLowAvoid hype-only

Score shortlist options; pick two, not six.

Sample 30/60/90-Day Executive Check-In Questions

  • Day 30: Are pilots live with real users? What friction surprised us?
  • Day 60: What do metrics say about quality and time saved?
  • Day 90: Scale, iterate, or stop—what does evidence support?

Key Takeaway

  • Tie every initiative to measurable outcomes and explicit non-goals.
  • Business cases need ranges and hidden costs (change management, monitoring, exit).
  • Pilots with kill switches beat permanent “experiments” with no owners.
  • Trust and training multiply value; tools alone do not.
  • Governance-lite still needs named roles, data classes, and review dates.