AI for directors is less about prompts and more about practice: turning scattered AI experiments into consistent, safe, measurable habits across a department. Executives set the direction and frontline teams do the work. Directors are where AI either becomes routine or stalls.
This playbook covers why AI has become a director-level responsibility, the five moves that make adoption stick, what to automate first, and a 90-day rollout plan.
Why Is AI a Director-Level Problem in 2026?
Because the gap between using AI and getting value from it now sits in the middle of the organization. BCG's 2026 AI at Work survey found:
- 74% of frontline employees use AI regularly, but 66% get little or no guidance on what to do with the time it saves.
- 72% say the skills expected of them have shifted, yet only 36% feel adequately upskilled.
- Integration of AI agents into workflows more than doubled in a year, to 30%, while half of employees report unclear governance.
- Only 28% of frontline employees see a strong connection between what leaders say about AI and what the organization actually does.
That last number is the director's opening. You can close the gap between what leadership says and what happens day to day.
The Director's AI Playbook: Five Moves
1. Use it yourself first
Credibility comes from first-hand use. Build two or three workflows on your own work before you ask anyone else to change theirs. Good starting points for directors:
- A portfolio status rollup from your teams' weekly updates
- A budget variance narrative from your monthly budget-vs-actuals export
- A vendor scoring pass against your weighted requirements
2. Pick three department workflows
Choose tasks that are repetitive, text- or data-heavy, and low risk if a draft is wrong. Reporting, documentation, and analysis usually qualify. Customer-facing decisions and anything involving personal data usually don't, at least not first.
3. Write one page of guardrails
Approved tools, data that never goes into a prompt, review rules, and who to ask when unsure. One page your managers can repeat from memory beats a policy nobody opens.
4. Name owners and measure
Give each workflow an owner and track a few honest numbers: hours saved, cycle time, and errors caught in review. Then decide, out loud, where the saved time goes.
5. Train by role, not by tool
Managers need status, 1:1, and metrics workflows. Analysts need SQL, cleaning, and documentation help. Generic "intro to AI" sessions rarely change behavior. Role-specific practice on real work does.
What Should Directors Automate First?
| Workflow | Input | Output | Risk if a draft is wrong |
|---|---|---|---|
| Portfolio rollup | Team status files | One-page RAG summary and escalations | Low: you review before sharing |
| Budget variance notes | Budget-vs-actuals export | Plain-English explanation of variances | Low: finance reviews |
| Vendor evaluation | Proposals and requirements | Weighted scoring and a draft memo | Medium: verify key claims |
| Executive update | Your notes | Three-paragraph update for leadership | Low |
| Interview kits | Job description | Structured questions and a scoring guide | Medium: keep decisions with people |
| Team AI playbook | Your guardrails and examples | A one-page guide with starter prompts | Low |
A sample prompt for the rollup, in Claude Code:
"Read the status files in the projects folder. Build a one-page portfolio rollup: RAG status, slipped milestones, and escalations needing my decision. List the source file for each item."
How Do You Measure AI Value at the Department Level?
Start small and specific. MIT's Project NANDA research found that 95% of organizations saw zero measurable return from generative AI pilots, and traced it to a learning gap, not the technology. Departments that measure a few workflows well, and keep improving them, avoid becoming part of that statistic.
Track three things per workflow: time per cycle before and after, the share of outputs that needed major edits, and what the team did with the time saved.
A 90-Day AI Rollout Plan for Directors
| Weeks | Focus | Outcome |
|---|---|---|
| 1–2 | Your own workflows | Two workflows running on your work |
| 3–4 | Guardrails and pilots | One-page guide; three pilot workflows with owners |
| 5–8 | Role-based enablement | Managers and analysts trained on their own workflows |
| 9–12 | Measure and expand | Results reviewed; next three workflows chosen |
Want a coach for the first 30 days? 1:1 Claude Code coaching for directors covers your own workflows, your department's highest-value process, and your team playbook across three sessions. $675 for the 3-session pack. Book a free 20-minute call.
Frequently Asked Questions
Q: How should directors start using AI? A: Start with your own recurring work: a portfolio rollup, a budget variance narrative, or a vendor comparison. Build two or three workflows yourself before rolling AI out to your teams, so your guidance comes from experience.
Q: Do directors need technical skills to use Claude Code? A: No. Claude Code takes plain-English instructions and asks permission before it changes anything. Directors with no coding background routinely build reporting and analysis workflows in their first session.
Q: How do I create an AI policy for my department? A: Write one page: approved tools, data that never goes into prompts, review rules, and who to ask. Add a short list of approved prompts for your department's real tasks. Keep it short enough that managers can repeat it from memory.
Q: How do I measure AI ROI in my department? A: Pick a few workflows and track time per cycle, the share of outputs needing major edits, and where saved time is reinvested. Small, honest measurements beat large estimates.
Q: Do you offer 1:1 AI coaching for directors? A: Yes. I coach directors and department heads 1:1 on Claude Code, ChatGPT, Codex, and Gemini, on their own work. Sessions are $250, or $675 for three, on Google Meet, with a 15-minute satisfaction guarantee.
