Implementation Guide: Generate lesson plans, curriculum materials, and differentiated learning resources

Hardware Procurement

Teacher Chromebook (Budget Option)

External Monitor for Chromeboxes

Wireless Access Point

Network Switch

Cloud-Managed Firewall/Gateway

Software Procurement

Prerequisites

• Active identity provider (Google Workspace for Education, Microsoft Entra ID, or Clever) with all teaching staff accounts provisioned and SSO configured
• Functioning Learning Management System (Canvas, Google Classroom, Schoology, or Microsoft Teams for Education) with current class rosters imported from SIS
• Student Information System (PowerSchool, Infinite Campus, Skyward, or Tyler SIS) with current enrollment data and OneRoster 1.2 or Clever integration enabled for roster sync
• Internet connectivity minimum 100 Mbps downstream / 50 Mbps upstream with <100ms latency to cloud services; recommended 500 Mbps+ symmetrical
• CIPA-compliant content filtering solution in place (GoGuardian, Lightspeed, Securly, or Cisco Umbrella) with ability to whitelist specific domains
• All teacher endpoint devices running a modern browser (Chrome 90+, Edge 90+, or Safari 15+); Chromebooks with ChromeOS 120+ recommended
• Cloud storage provisioned for all teaching staff: Google Drive (15GB+ free) or OneDrive (1TB with M365 A3)
• Administrative access to Google Admin Console or Microsoft 365 Admin Center for SSO app registration, domain whitelisting, and user group management
• Identified project champion(s): 1-2 tech-savvy teachers or curriculum coordinators who will pilot the tools and provide peer support
• Completed vendor Data Privacy Agreement (DPA) review process: identify the school/district DPA signatory and legal review workflow before procurement
• Current curriculum standards documentation: state standards, pacing guides, and any existing curriculum frameworks that AI tools will align to
• Network firewall rules allow outbound HTTPS (port 443) to: *.magicschool.ai, *.eduaide.ai, *.canva.com, *.openai.com, *.googleapis.com, *.microsoft.com, and associated CDN domains

Installation Steps

...

Step 1: Environment Assessment & Documentation

Conduct a thorough audit of the client's existing technology infrastructure. Document the current LMS, SIS, identity provider, productivity suite, network topology, content filtering solution, and any existing AI tools in use. Identify the client's curriculum standards (Common Core, state-specific, NGSS, etc.) and current lesson planning workflow. Catalog all teaching staff who will need access, including their roles (classroom teacher, specialist, curriculum coordinator, administrator). Create a network diagram showing internet connectivity, firewall, content filter, and wireless infrastructure.

chrome://version

Step 2: Compliance & Data Privacy Agreement Execution

Before procuring or deploying any AI platform, execute the required Data Privacy Agreements (DPAs) between the education client and each AI vendor. This is a legal requirement under FERPA and most state student data privacy laws (e.g., NY ED Law 2-d, CA SOPIPA). Work with the school/district's designated DPA signatory (typically the superintendent, principal, or technology director). Verify each vendor's compliance certifications (SOC 2 Type II, FERPA attestation, COPPA compliance). For COPPA specifically, note the new rules effective June 23, 2025, with full compliance required by April 22, 2026.

# DPA Tracking Spreadsheet Fields:
# Vendor | DPA Date | Expiry | Signatory | Data Categories | Compliance Certs

Step 3: Content Filter Configuration & Domain Whitelisting

Configure the school's content filtering solution to whitelist all AI platform domains while maintaining CIPA-compliant filtering. Create a dedicated URL allowlist category for 'Approved AI Tools' to simplify future management. Ensure that AI tool domains are accessible from both the staff network/VLAN and any student-accessible networks where AI tools will be used under teacher supervision. ``` # GoGuardian Admin - Add URL Allowlist entries # Navigate to: GoGuardian Admin Console > Policies > ...

Step 4: Identity Provider & SSO Configuration

Configure Single Sign-On (SSO) for all AI platforms through the school's identity provider. The preferred path for education is Clever SSO (used by 80%+ of US K-12 districts), which provides one-click login and automatic roster sync. For schools not on Clever, configure direct SAML/OIDC SSO through Google Workspace or Microsoft Entra ID. This step ensures teachers can access AI tools with their existing school credentials and eliminates separate password management.

Option A: Clever SSO (Recommended for K-12)

Option B: Google Workspace SAML SSO

Option C: Microsoft Entra ID SSO

Verify SSO Login

Step 5: MagicSchool AI Enterprise Deployment

Deploy and configure MagicSchool AI as the primary content generation platform. This includes admin console setup, organizational structure, feature configuration, guardrail settings, and initial teacher account provisioning. MagicSchool Enterprise provides district-level admin controls including usage analytics, PII detection guardrails, custom prompt templates, and role-based access.

1. Admin Console Initial Setup

2. Configure Guardrails & Safety Settings

3. Configure Available Tools

4. User Provisioning (if not via Clever/SSO auto-provision)

5. Verify Deployment

Step 6: Eduaide.ai Pro Deployment

Deploy Eduaide.ai as a complementary content generation platform. Eduaide excels at differentiated materials, graphic organizers, and game-based learning activities, filling gaps not fully covered by MagicSchool. For smaller clients (tutoring centers), Eduaide may serve as the primary platform due to its lower per-seat cost.

Step 7: Canva for Education Setup & LMS Integration

Register the school for Canva for Education (100% free for eligible K-12 institutions) and configure LMS integration for seamless content sharing. Canva provides AI-powered visual content creation including Magic Write (text generation), Magic Design (template suggestions), and thousands of education-specific templates for lesson materials, presentations, worksheets, and infographics.

Configure LMS Integration — Canvas LTI

Config URL: https://www.canva.com/lti/configure

Step 8: AI Add-on Deployment (Google AI Pro or Microsoft Copilot)

Deploy the native AI add-on for the client's primary productivity suite. This embeds AI content generation directly into the tools teachers use daily (Google Docs/Slides or Word/PowerPoint), complementing the purpose-built education AI platforms. Choose based on the client's existing ecosystem: Google AI Pro for Education (Google shops) or Microsoft 365 Copilot for Education (M365 shops).

Option A: Google AI Pro for Education

Option B: Microsoft 365 Copilot for Education

# Connect to Microsoft Graph PowerShell
Connect-MgGraph -Scopes 'User.ReadWrite.All','Organization.Read.All'

# Get the Copilot for Education license SKU
$sku = Get-MgSubscribedSku | Where-Object { $_.SkuPartNumber -like '*COPILOT*EDU*' }

# Assign to a group of users (requires group ID)
# Use Microsoft 365 Admin Center group-based licensing for simplicity

Step 9: Custom Prompt Template Library Deployment

Create and deploy a curated library of education-specific prompt templates tailored to the client's curriculum standards, grade levels, and instructional frameworks. This is a high-value MSP deliverable that dramatically improves AI output quality and reduces the learning curve for teachers. Templates should be organized by subject, grade band, and resource type, and stored in a shared location accessible to all teaching staff. ``` # 1. Create shared Google Drive folder (or SharePoint library) ...

Step 10: LMS Integration & Content Distribution Workflow

Configure the end-to-end workflow from AI content generation through LMS distribution. This ensures generated lesson plans and materials flow seamlessly from AI tools into the LMS where students and teachers access them. Set up the export-to-LMS pipeline for each AI platform, configure assignment templates, and verify the complete workflow from generation to student delivery. ``` # === Google Classroom Integration === # MagicSchool > Generate lesson plan > Click 'Export' # Select: 'Google Docs'...

Step 11: Network Optimization & Monitoring

Configure network monitoring to track AI tool performance and bandwidth consumption. Optimize QoS settings to prioritize educational SaaS traffic during school hours. Set up alerts for connectivity issues that would impact AI tool availability. This step ensures the AI platform SLAs are met from the client's network perspective.

UniFi Network Optimization (if Ubiquiti)

Traffic Management

Uptime Monitoring

Uptime Robot Monitor Configuration

• https://app.magicschool.ai — HTTP 200 check, 5-min interval
• https://app.eduaide.ai — HTTP 200 check, 5-min interval
• https://www.canva.com — HTTP 200 check, 5-min interval
• Alert method: Email + Slack webhook to MSP NOC

Chromebook Network Diagnostics

# run on teacher Chromebook to verify DNS and HTTPS connectivity

chrome://network

Step 12: Teacher Training & Pilot Launch

Conduct structured teacher training sessions and launch the pilot program with 5-10 teacher champions before full rollout. Training should cover: AI tool operation, effective prompt crafting, output review/editing best practices, compliance guardrails (no PII in prompts), differentiation workflows, and LMS integration. Provide ongoing support channels for the pilot period (2-4 weeks) before expanding to all staff.

Training Session Structure (2.5 hours recommended)

Module 1: AI Fundamentals for Educators (30 min)

• What generative AI can and cannot do
• How AI content generation works (no jargon)
• The human-in-the-loop principle: AI drafts, teacher refines
• Compliance rules: NEVER put student names, IDs, or PII into AI tools

Module 2: MagicSchool Hands-On Workshop (45 min)

Module 3: Prompt Crafting Best Practices (30 min)

• Show prompt template library
• Practice: Generic prompt vs. detailed prompt comparison
• Framework: Subject + Grade + Standard + Duration + Special Needs
• Demonstrate differentiation: same topic, 3 reading levels

Module 4: Workflow Integration (30 min)

• End-to-end: AI generate > Review > Edit > LMS publish
• Canva for visual materials
• Saving and organizing generated content
• Sharing effective prompts with colleagues

Module 5: Q&A and Pilot Kickoff (15 min)

Pilot Feedback Form — Google Forms Question Set

Custom AI Components

5E Lesson Plan Generator

Type: prompt A structured prompt template that generates complete lesson plans using the 5E instructional model (Engage, Explore, Explain, Elaborate, Evaluate). This model is widely adopted in K-12 science education and increasingly in other subjects. The prompt produces standards-aligned, time-blocked lesson plans with built-in differentiation and formative assessment checkpoints. Designed for direct use in MagicSchool custom tools, Eduaide, or as an OpenAI API prompt. Implementation:...

Differentiated Content Leveler

Type: prompt A prompt template that takes any text passage, article, or content piece and generates three differentiated versions: Below Grade Level (with scaffolding), On Grade Level (standard), and Above Grade Level (enriched). This is critical for inclusive classrooms serving students with diverse reading abilities, ELL students, and gifted learners. The output includes Lexile-approximate levels and comprehension questions at each level.

Implementation

System Prompt

You are a reading specialist and curriculum differentiation expert. You transform educational texts into multiple reading levels while preserving core content and learning objectives. You follow research-based reading level guidelines and include appropriate scaffolding for each level. You understand Lexile levels, Flesch-Kincaid readability, and Fountas & Pinnell guided reading levels. Never include real student names or PII.

User Prompt Template

# fill in all {{variables}} before sending

Differentiate the following content into THREE reading levels. Maintain the same core concepts and key vocabulary at all levels.

**Source Content:**
{{content}}

**Subject:** {{subject}}
**Target Grade Level:** {{grade_level}}
**Content Standard:** {{standard}}
**Key Vocabulary to Preserve:** {{vocabulary}}

VERSION 1: Below Grade Level (Approaching)

• Target: 1-2 grade levels below {{grade_level}}
• Approximate Lexile: {{below_lexile}}

Modifications:

• Shorter sentences (8-12 words average)
• High-frequency vocabulary with context clues built in
• Bolded key vocabulary with inline definitions
• Chunked paragraphs (2-3 sentences each)
• Subheadings added for text navigation
• Visual cue suggestions in [brackets]

Scaffolding Supports:

• Word bank with definitions
• Sentence frames for response questions
• Graphic organizer suggestion

Comprehension Questions (3):

• 1 literal/recall question
• 1 inference question with sentence stem
• 1 text evidence question with page/paragraph reference

VERSION 2: On Grade Level (Meeting)

• Target: {{grade_level}} level
• Approximate Lexile: {{on_lexile}}

Features:

• Grade-appropriate sentence complexity
• Content-specific vocabulary in context
• Standard paragraph structure
• Text features (headings, captions if applicable)

Comprehension Questions (4):

• 1 literal/recall question
• 1 inference question
• 1 text evidence question
• 1 analysis/evaluation question

VERSION 3: Above Grade Level (Exceeding)

• Target: 1-2 grade levels above {{grade_level}}
• Approximate Lexile: {{above_lexile}}

Enhancements:

• Complex sentence structures with embedded clauses
• Advanced academic vocabulary with etymological notes
• Extended content with additional detail/nuance
• Primary source excerpts or data (if applicable)
• Cross-curricular connections noted

Comprehension Questions (4):

• 1 analysis question
• 1 synthesis question connecting to prior knowledge
• 1 evaluation question requiring argumentation
• 1 open-ended inquiry question for independent research

Teacher Notes

• Suggested grouping strategy for using all three levels
• How to use these levels in a jigsaw or literature circle format
• Assessment alignment: how to evaluate all levels against the same standard

Lexile Reference Table (Parameter Defaults)

{
  "grade_2": {"below": "300-400L", "on": "420-650L", "above": "650-800L"},
  "grade_3": {"below": "400-500L", "on": "520-820L", "above": "820-1000L"},
  "grade_4": {"below": "500-600L", "on": "740-940L", "above": "940-1100L"},
  "grade_5": {"below": "600-750L", "on": "830-1010L", "above": "1010-1200L"},
  "grade_6": {"below": "750-850L", "on": "925-1070L", "above": "1070-1250L"},
  "grade_7": {"below": "850-950L", "on": "970-1120L", "above": "1120-1300L"},
  "grade_8": {"below": "900-1000L", "on": "1010-1185L", "above": "1185-1385L"}
}

MagicSchool Configuration

• Use built-in "Text Rewriter / Leveler" tool as starting point
• Supplement with this custom prompt for more detailed differentiation
• Input: Paste source text, select grade level, click generate

API Implementation (Python)

def differentiate_content(
    content: str,
    subject: str,
    grade_level: int,
    standard: str,
    vocabulary: list[str],
    model: str = "gpt-5.4-mini"
) -> str:
    lexile_map = {
        2: ("300-400L", "420-650L", "650-800L"),
        3: ("400-500L", "520-820L", "820-1000L"),
        4: ("500-600L", "740-940L", "940-1100L"),
        5: ("600-750L", "830-1010L", "1010-1200L"),
        6: ("750-850L", "925-1070L", "1070-1250L"),
        7: ("850-950L", "970-1120L", "1120-1300L"),
        8: ("900-1000L", "1010-1185L", "1185-1385L"),
    }
    below, on, above = lexile_map.get(grade_level, ("700-900L", "900-1100L", "1100-1300L"))
    vocab_str = ", ".join(vocabulary)
    
    # Build prompt with template above, substituting variables
    # Call OpenAI API with temperature=0.6 for consistent leveling
    # Return structured output with all three versions
    pass

Bulk Curriculum Generation Workflow

Type: workflow An automated workflow for generating an entire unit or semester of curriculum materials in bulk. Uses the OpenAI Batch API for 50% cost savings on large generation jobs. Takes a curriculum scope-and-sequence document as input and produces lesson plans, assessments, vocabulary lists, and differentiated materials for every lesson in the unit. Designed for curriculum coordinators or MSP-managed content generation services. Implementation: ``` ## Bulk Curriculum Generation ...

PII Detection Guard

Type: skill A pre-processing guard that scans teacher prompts for personally identifiable information (PII) before sending to any AI platform. Detects student names, ID numbers, email addresses, phone numbers, dates of birth, and other FERPA-protected data. Returns a warning and sanitized prompt. This is a defense-in-depth measure complementing MagicSchool's built-in PII detection. Implementation: ``` ## PII Detection Guard ### Purpose Provides a client-side PII detection layer for c...

Curriculum Standards Alignment Checker

Type: prompt A quality assurance prompt that takes any AI-generated lesson plan or curriculum material and verifies its alignment to specific state or national standards. Returns a detailed alignment report with coverage analysis, identifying which standards are fully addressed, partially addressed, or missing. Essential for curriculum coordinators reviewing AI-generated content before adoption. Implementation: ``` ## Curriculum Standards Alignment Checker ### System Prompt: You are ...

Parent Communication Generator

Type: prompt Generates parent-facing communications about classroom learning in clear, accessible language. Produces weekly newsletters, unit overview letters, and progress update templates in multiple languages. Designed to bridge the home-school communication gap and increase parent engagement. Outputs are culturally sensitive and available at multiple reading levels. Implementation: ``` ## Parent Communication Generator ### System Prompt: You are a family engagement specialist who...

Testing & Validation

Client Handoff

The client handoff meeting should be a 60-90 minute session with the school/district technology director, curriculum coordinator, and 2-3 teacher champions present. Cover the following:

1. Solution Overview & Architecture (15 min)

• Walk through the deployed solution: MagicSchool AI (primary), Eduaide.ai (complementary), Canva for Education (visual content), and native AI add-on (Google AI Pro or Microsoft Copilot)
• Show the integration architecture: SSO → AI Platform → Export → LMS → Students
• Demonstrate the admin consoles and where to find usage analytics

2. Compliance Documentation Review (15 min)

• Hand over the compliance folder containing: all executed DPAs, vendor SOC 2 reports, content filter configuration documentation, PII handling policy, data flow diagram, and parental consent templates (if applicable)
• Review DPA renewal dates and set calendar reminders
• Explain the teacher PII training materials and quiz

3. Teacher Training Materials (10 min)

• Hand over: Quick Start one-pager (login URL, top 5 prompts, export steps, 'never do' list), recorded training video, prompt template library (shared Google Drive folder), and FAQ document
• Show the teacher support channel (Google Chat space or Teams channel)
• Identify the internal 'AI Champion' teacher(s) for peer support

4. Live Demonstration (20 min)

• Generate a lesson plan end-to-end: open MagicSchool → select tool → enter parameters → generate → review → export to Google Docs → create LMS assignment
• Show differentiation workflow: same content → three reading levels in Eduaide
• Show visual content creation in Canva for Education
• Demonstrate the admin dashboard: usage metrics, PII detection logs, user management

5. Success Criteria Review (10 min)

• Review agreed-upon success metrics: teacher adoption rate target (>80% within 60 days), average time savings per teacher (target: 5+ hours/week), content quality rating from pilot (target: 4/5+), zero PII incidents, compliance audit readiness
• Schedule 30-day and 90-day review meetings
• Define escalation path: teacher → AI Champion → school tech contact → MSP help desk

6. Ongoing Support & SLA (10 min)

• Review the MSP support agreement: response times, included services, out-of-scope items
• Provide MSP contact information: help desk email/phone, account manager, emergency contact
• Explain the monthly optimization review process: usage reporting, prompt library updates, vendor update notifications

Documentation to Leave Behind

Maintenance

Monthly Maintenance Tasks (2-4 hours/month)

Quarterly Maintenance Tasks

Annual Maintenance Tasks

SLA Considerations

• AI platform availability is dependent on vendor SLAs (MagicSchool, Eduaide, Canva all cloud-hosted).
• MSP SLA covers: SSO/authentication issues (4-hour response), content filter/network issues (2-hour response during school hours), user provisioning (24-hour turnaround), and compliance incidents (immediate response).
• Critical path: If MagicSchool goes down, teachers can use Eduaide.ai or direct Google AI/Copilot as fallback.
• If internet goes down, no AI tools are available—this is the single point of failure, emphasizing the importance of reliable network infrastructure.

Escalation Path

Alternatives

Microsoft Copilot-Centric Approach

Instead of purpose-built education AI platforms (MagicSchool/Eduaide), rely entirely on Microsoft 365 Copilot for Education embedded in Word, PowerPoint, OneNote, and Teams. Teachers use Copilot within their existing M365 productivity tools to generate lesson plans, presentations, and assessments without switching to a separate AI platform.

Google AI Pro-Centric Approach

For Chromebook-heavy schools already on Google Workspace for Education, rely on Google AI Pro for Education (Gemini in Workspace) plus Canva for Education as the primary content generation stack, supplemented by free tools like Khanmigo. Avoids any additional paid AI platform subscriptions.

Open Source Self-Hosted LLM

Deploy an open-source LLM (Llama 3.1 8B or 70B) on-premises for maximum data sovereignty and zero per-API-call costs. Build a custom web interface for teachers to access the model. All data stays within the school's network—no student-adjacent data ever leaves the building.

Eduaide.ai as Sole Platform (Budget Approach)

Use Eduaide.ai Pro ($5.99/month per teacher) as the sole paid AI content generation platform, supplemented by free tools: Canva for Education, Khanmigo, and free tiers of MagicSchool and Google Gemini. Minimizes licensing costs for small tutoring centers or individual schools with tight budgets.

Custom API-First Approach

Skip purpose-built education platforms entirely and build a custom content generation solution using OpenAI GPT-5.4 mini or Anthropic Claude APIs. Develop a custom web application with education-specific prompt templates, integrate directly with the school's LMS via API, and maintain full control over the user experience and data flow.