Moving beyond single prompts to building an AI-powered team requires understanding three key concepts: Personas, Skills, and Subagents. Together, they transform AI from a simple question-answering tool into a coordinated system that amplifies your capabilities.
Personas
Assign personas to generate customized perspectives.
A persona is a role you give AI to shape its responses. When you tell AI "You are an experienced curriculum developer," it adjusts its tone, focus, and recommendations accordingly.
How Personas Work
Personas provide context that shapes AI behavior:
- Role definition: "You are a veteran high school principal with 20 years of experience"
- Expertise framing: "You specialize in student data privacy and FERPA compliance"
- Perspective setting: "You are a parent reviewing this communication"
Practical Examples for Educators
| Persona | Use Case | | ----------------------------- | ----------------------------------------------------------------------------- | | "Innovative CBO" | Budget presentations that balance fiscal responsibility with strategic vision | | "Curriculum Developer" | Lesson plans aligned with standards and pedagogical best practices | | "Parent Advocate" | Review communications for clarity and tone from a family perspective | | "Board Member" | Anticipate questions and concerns about proposed policies | | "Student Services Specialist" | IEP documentation and intervention recommendations |
Setting Up Personas
In Claude Projects or ChatGPT Custom GPTs:
Create a project with custom instructions that establish the persona. Every conversation in that project starts with the persona already active.
Example Custom Instruction:
You are an experienced K-12 curriculum coordinator with deep expertise in differentiated instruction and Universal Design for Learning. When reviewing lesson plans, you always consider diverse learner needs. You communicate in a collegial, supportive tone appropriate for professional development settings.
Skills
Teach AI specific workflows, formats, or domain knowledge.
Skills are reusable prompts or instructions that standardize how AI handles specific tasks. They ensure consistency across multiple uses and save time by encoding your preferences.
How Skills Work
Skills capture:
- Formats: "Always structure meeting notes with attendees, decisions, action items, and next steps"
- Workflows: "When analyzing data, first summarize key findings, then identify trends, then recommend actions"
- Domain knowledge: "Our district uses these specific terms for..."
- Quality standards: "Communications should be at an 8th-grade reading level"
Practical Examples for Educators
Lesson Plan Generator Skill:
When creating a lesson plan, always include: learning objectives aligned to standards, estimated time, materials needed, differentiation strategies for ELL and special education students, formative assessment checks, and extension activities.
Board Report Skill:
Format board reports with: executive summary (3 sentences max), background context, data/evidence, recommendation, fiscal impact, and next steps. Use formal but accessible language.
Parent Communication Skill:
Write parent communications that are warm but professional. Lead with the most important information. Include specific dates and action items. Keep to 200 words maximum. Close with contact information for questions.
Building Your Skill Library
Start documenting effective prompts:
- Notice what works: When AI produces something great, save the prompt
- Standardize formats: Create templates for recurring tasks
- Share with colleagues: Build a team library of proven skills
- Iterate: Refine skills based on results
Subagents
Delegate specific tasks to specialized AI instances.
Subagents are AI assistants configured for specific jobs. Rather than one generalist AI, you have a team of specialists you can call on for different tasks.
How Subagents Work
Each subagent has:
- Specific purpose: "Research this topic, then summarize findings"
- Defined scope: Focused on one type of task
- Consistent behavior: Same approach every time
- Potential autonomy: Can complete multi-step tasks independently
Practical Examples for Educators
Assessment Writer Subagent:
Purpose: Create quizzes and assessments aligned to learning objectives Scope: Multiple choice, short answer, rubric-based items Behavior: Always includes answer key, difficulty levels, and standards alignment
Research Summarizer Subagent:
Purpose: Digest academic research for practical application Scope: Education research, policy briefs, best practices Behavior: Extracts key findings, identifies implementation considerations, notes limitations
Meeting Prep Subagent:
Purpose: Prepare for meetings by gathering relevant context Scope: Board meetings, IEP meetings, parent conferences Behavior: Creates agenda, anticipates questions, suggests talking points
Creating Subagents
Using Claude Projects:
- Create a new project
- Add custom instructions defining the subagent's role and behavior
- Upload relevant documents (templates, standards, examples)
- Name the project clearly (e.g., "Assessment Writer")
Using ChatGPT GPTs:
- Create a custom GPT
- Define instructions, capabilities, and conversation starters
- Configure knowledge files
- Share with team if needed
Integrations and MCPs
Connect AI to external tools and data sources.
The most powerful AI setups connect to your existing systems. Model Context Protocol (MCP) and other integrations allow AI to:
- Access your documents in Google Drive or OneDrive
- Query your databases
- Connect to your communication tools
- Interact with your learning management system
Examples
- "Check our Google Drive for the latest policy template"
- "Pull last month's attendance data from our SIS"
- "Schedule this meeting using my calendar"
Current State
Integrations are evolving rapidly. Start with platform-native features (file uploads, code interpreter) before exploring advanced integrations. For most education use cases, uploading relevant documents provides sufficient context.
How They Work Together
The real power emerges when personas, skills, and subagents combine:
Scenario: End-of-Year Report Preparation
- Subagent (Data Analyst) pulls and summarizes key metrics
- Skill (Board Report Format) structures the output correctly
- Persona (Superintendent) reviews for executive-level communication
Scenario: Curriculum Development
- Persona (Curriculum Expert) guides alignment decisions
- Skill (Lesson Plan Generator) ensures consistent format
- Subagent (Assessment Writer) creates aligned assessments
Getting Started
Week 1: Experiment with Personas
Try assigning different roles to AI and notice how responses change.
Week 2: Document Your First Skill
Capture an effective prompt you use repeatedly. Refine it.
Week 3: Create Your First Subagent
Set up a Claude Project or Custom GPT for one specific task.
Ongoing: Build Your Team
Gradually add more specialized assistants. Share what works with colleagues.
The Key Insight
The difference between casual AI users and power users often comes down to this: casual users ask AI questions; power users build AI teams.
You're not just using a chatbot. You're assembling an AI-powered workforce tailored to your specific needs.
When done well, this approach scales your capacity without scaling your workload. Each persona, skill, and subagent you develop becomes a reusable asset that pays dividends across all future work.