Transition Guide — Employees

Operational Policy — Employees

This document applies the principles of the AI Execution Standards to your role transition. The same rigor the organization demands for delegating work to AI is the rigor expected in your own transformation.

The operating principle is the Universal Translation Rule: replace "the human produces the artifact" with "the human defines the spec → the system produces the artifact." Your transition consists of applying this principle to your own role.

1. Guiding Principle

You are responsible for the transformation of your role.

Your manager provides the framework and support. You provide the plan and the execution.

The expected outcome: your role operates at Level 2 (AI-Integrated). AI is no longer a tool you choose to use — it's an integral part of how you work. For engineering, the target is Level 3 (AI-Native).

2. Transformation Layers

Your transition has four layers. Each is a prerequisite for the next.

Layer 1 — AI Literacy (prerequisite)

You must:

Understand what the available AI tools can do (and can't do)
Know how to use at least one AI tool in a structured way (not just ad hoc prompts)
Understand the difference between using AI like Google (Level 1) and integrating AI into a workflow (Level 2)

Minimum bar: You can explain, for your role, 3 tasks where AI delivers a measurable gain.

Layer 2 — Mapping Your Actual Work

Before redesigning your role, document it honestly:

What you actually do on a daily basis (not your job description)
How much time each task category takes
Which tasks are repetitive, predictable, or template-based
Which tasks require human judgment
Where your bottlenecks are

Requirement: No redesign without mapping. You can't transform what you don't understand.

Layer 3 — Role Redesign

For each task category identified in Layer 2, answer:

Should this task exist if AI is available?
If yes, which part is human (judgment, decision, relationship) and which part is system (execution, generation, verification)?
What AI tool or workflow would replace the system part?
What would the measurable gain be (time, quality, volume)?

Expected result: a clear vision of your redesigned role, not a wish list.

Layer 4 — Implementation (highest standard)

You move from plan to action:

Build or configure the systems identified in Layer 3
Integrate AI into your daily workflows
Measure results
Iterate

Rule: If you can't show a workflow that has changed, you haven't implemented.

2b. Brief Primitives (skills to demonstrate)

The transition brief you deliver is built from six primitives. Each is a skill.

Primitive 1 — Honest Description of Current Role

Describe your role as it actually operates. Not your title, not your aspirations — what you do.

Exercise: For one week, log every task and the time spent. Group by category. That's your starting point.

Primitive 2 — AI-First Vision of the Role

Imagine your role is being recruited for today at an AI-native company. What would the job description look like? Which tasks would no longer exist? What new skills would be required?

Primitive 3 — Gap Analysis

Compare Primitive 1 and Primitive 2. The gap is your work plan. Be specific: "I manually write weekly reports in 3 hours" → "An AI workflow could generate them in 20 minutes from existing data."

Primitive 4 — System Design

For each identified gap, specify:

What system or automation fills that gap
What tools are needed
What you can build on your own vs what needs support
Estimated effort

Primitive 5 — Metrics Design

For each system, define how you'll know it's working. Numbers, not feelings: time saved, volume increased, errors reduced, turnaround shortened.

Primitive 6 — 30/60/90 Day Plan

Break your transformation into three horizons:

30 days: Layers 1-2 completed. You understand your current role and the AI possibilities. At least 1 AI workflow in use.
60 days: Layer 3 completed. Role redesigned. 2-3 workflows transformed and measured.
90 days: Layer 4 underway. Level 2 demonstrated. Measurable evidence of gains.

3. Brief Quality Rules

A valid brief must be:

Honest — Based on your actual work, not on your perception of what leadership wants to hear.

Specific — "I'll use AI to be more efficient" is not a plan. "I'll replace manual client report writing with a Claude workflow that generates the report from CRM data" is a plan.

Testable — Your manager (or a peer) can evaluate your progress without asking you for explanations.

Measured — Every improvement has an associated metric.

4. Readiness Checklist

Before submitting your brief, confirm:

I've documented my actual work (not my job description)
I've identified at least 3 tasks that can be transformed by AI
I can explain what Level 2 means for my specific role
My plan contains concrete systems to build, not vague intentions
Every improvement has a measurable metric
My 30/60/90 plan has verifiable deliverables at each stage

If any answer = no → your brief isn't ready. Ask your manager for help.

5. Evaluation Standard

Your progress is evaluated on evidence:

What you've built (systems, workflows, automations)
What has improved because of you (before/after metrics)
What exists now that didn't exist before
Evidence required. Narrative isn't enough.

What is NOT evaluated:

How many times you've used ChatGPT
How many hours spent "learning AI"
Your enthusiasm or attitude toward change

6. Diagnosis: if you're stuck

If your transition stalls, diagnose the layer — and ask for help:

Symptom	Probable cause	Action
I don't know where to start	Insufficient AI literacy (Layer 1)	Ask your manager for training or mentoring
I don't know what to transform	Missing mapping (Layer 2)	Do the one-week documentation exercise
I know what to transform but not how	Incomplete redesign (Layer 3)	Work with your manager or a technical peer
I have a plan but nothing is changing	Missing implementation (Layer 4)	Start with the simplest workflow, not the most ambitious

Identify the layer, ask for the appropriate support. Staying stuck in silence is not expected — asking for help is.

7. Support Provided

The organization is actively investing in your transition:

Tools: Access to AI licenses based on your role (see list below). If you're missing a tool, ask for it — don't stay stuck.
Time: Protected time for learning and experimentation, to be negotiated with your manager. Experimentation is part of the job, not an extra.
Coaching: Your manager is responsible for providing you with context, constraints, and feedback on your brief. If you're not getting that support, escalate.
Peers: Share your discoveries, prompts, and workflows with your colleagues. Transformation is easier as a team.
AI clinics: regular sessions (organized by your manager) where the team shares discoveries and blockers. Short format, concrete learning.
Transition pairing: if you're learning, ask to work with a more advanced colleague on a real case. If you're advanced, offer to share.
Peer brief review: before submitting, have a colleague review your brief. An outside perspective catches blind spots.
Escalation: If a systemic blocker is preventing your progress (missing tool, access denied, budget), report it to your manager who will escalate to leadership.

Approved tools (list maintained by leadership):

AI assistants: Claude, ChatGPT (organizational accounts)
Code: GitHub Copilot, Claude Code (engineering)
Automation: Zapier, n8n, Make (department-dependent)
Prohibited for security reasons: any tool that sends client or financial data to an unapproved third-party service
Requesting a new tool: talk to your manager with a business justification — response within 1 week

What the organization will not provide:

A ready-made plan for your role — you know your work better than anyone, and that expertise is what the plan is built on
A one-size-fits-all generic training — the transformation is specific to each role, and learning happens by building

8. Documentation Standard

Your brief and progress reports are specifications, not stories.

Your documents must:

State facts (not intentions)
Define the terms you use
Quantify results
Include constraints encountered
Be understandable by someone who doesn't know your role

A brief that says "I'm going to use AI to improve my work" without specifying what, how, and how to verify, is not a brief.

9. What is NOT expected

The transformation has limits. The goal isn't to put AI everywhere:

Don't automate what requires your judgment — sensitive client decisions, ambiguous situations, complex human relationships. That's where your value is irreplaceable.
Don't break a process that works — if a manual workflow is simple, fast, and reliable, forcing it into AI to prove your adoption isn't transformation, it's waste.
Don't sacrifice quality for speed — an AI workflow that produces mediocre content faster isn't a gain.
Don't be afraid to experiment and fail — a failed attempt that teaches you something has more value than no attempt at all. Experimentation is part of the process.

The rule: if AI improves the outcome or frees up your time for higher-value work, it's a good transformation. Otherwise, look elsewhere.

10. Common Pitfalls

The most common mistakes in this type of transition — recognize them to avoid them:

Using AI only as a search engine or grammar checker and considering that Level 2 — that's still Level 1
Submitting a vague brief with no metrics ("I'll integrate AI into my daily work") — that's an intention, not a plan
Waiting for your manager to tell you exactly what to do — you know your role better than anyone, start there
Confusing activity with results — using AI 4 hours a day without changing a workflow is not transformation
Postponing your transition waiting for "the right tool" or "the right training" — start with what's available, adjust later

11. Performance Expectation

You are evaluated on:

The quality of your brief (specificity, honesty, measurability)
Systems built or workflows transformed
Gains measured (not estimated — measured)
Your ability to identify and fill your own gaps
Your contribution to raising the team baseline (shared prompts, documented workflows)

Not on your enthusiasm. Not on your prompt volume. On what has changed.

Appendix A: What Level 2 concretely means

The descriptions below illustrate Level 2 by department. They're not exhaustive — they show the direction.

Engineering (target: Level 3)

Code generated and reviewed by AI, directed by the human
The human defines the architecture, constraints, and tests — AI produces and validates the code
Every engineer builds their own agent that delivers demonstrable value
Engineers in the AI Lab operate at Level 3: non-interactive development, specifications and scenarios drive autonomous agents

Marketing

Campaigns generated by AI from strategic briefs
A/B testing of AI-produced variants
Automated reporting
Content creation via structured workflows, not ad hoc prompts

Sales

Outreach and follow-ups generated by AI from qualification rules
Proposals produced by AI from templates and client data
CRM enriched by AI
Human time focused on relationships and judgment

Customer service

Standard responses generated by AI, validated by the human
Automated client health scoring
Escalation based on rules, not intuition
Knowledge base maintained by AI

Finance

Reports generated by AI from defined financial models
Automated anomaly detection
Forecasts produced by system
Human time focused on analysis and decisions

Appendix B: Minimum acceptable transformation by department

This is the floor — the smallest transformation that demonstrates Level 2. If you don't even reach this, your transition hasn't started.

Engineering: At least 1 recurring task (tests, review, documentation) delegated to an agent with a reproducible workflow.
Marketing: At least 1 content type produced via a structured AI workflow (brief → generation → revision), not ad hoc prompts.
Sales: At least 1 step of the sales cycle (outreach, follow-up, or proposal) generated by AI from documented rules.
Customer service: At least 1 category of standard responses generated by AI and validated by the human, with a reusable template.
Finance: At least 1 recurring report generated by AI from a defined model, instead of being built manually.

Appendix C: Brief example — good vs bad

Bad brief

"I'm going to use Claude to help me with my daily work. I plan to use it for writing emails, doing research, and improving my productivity. My plan: start using it more regularly and measure gains over time."

Why it's insufficient: no specific workflow, no metrics, no systems to build, no 30/60/90. It's a statement of intent, not a plan.

Good brief (example: customer service representative)

Current role: I handle ~40 tickets/day. ~60% are recurring questions (password reset, delivery status, refund requests). Average time per recurring ticket: 8 minutes. Complex tickets (escalations, technical cases): ~15/day, 20 minutes each.

AI-first role: Recurring tickets are generated by AI from templates and the knowledge base, validated by me in <2 minutes. My time focuses on complex cases and improving the knowledge base.

Gap: No structured templates. No AI workflow. Knowledge base incomplete and not formatted for AI.

Systems to build: (1) Response templates for the 10 most frequent ticket types. (2) Claude workflow: ticket → classification → draft response → human validation. (3) Knowledge base structured for AI.

Metrics: Time per recurring ticket (from 8min to <2min). Tickets handled per day (from 40 to 60+). Client satisfaction rate maintained or improved.

30/60/90: 30d — top 10 ticket types documented, 3 templates in use. 60d — Claude workflow in production for recurring tickets, metrics collected. 90d — 60%+ of recurring tickets handled via the workflow, knowledge base up to date.

Why it's a good brief: honest about the current state, specific about the systems, measurable, broken into verifiable steps.

Summary Rule

Your role isn't to use AI. Your role is to redefine how your work gets done because AI exists.

If you can't show what has changed, nothing has changed.

← Back to home · The reference framework · Manager guide · Glossary