AI-Driven Operations:
Automated Learning Support & Workflow Ecosystem

A SAM-based AI ecosystem designed to automate complex operational workflows and provide immediate user assistance. The system integrates a ① 24/7 multilingual AI chatbot for real-time operational support and an ② AI analytics agent for automated student progress reporting. By offloading recurring manual tasks to AI, Lead Tutors now save 2+ hours weekly, allowing them to focus entirely on direct student interaction and program quality.

Experience the Project

Audience: Operational Staff and Program Leadership at Seed Academy

Role: AI Systems Integration, Operational Workflow Automation, Learning Technology, Prompt Engineering, User Support Strategy

Tools & Frameworks: Chatbase, Zapier, GPT-4o, Gemini, Camtasia, Canva, Claude AI, SAM (Successive Approximation Model)

The Problem

SEED Academy, a Bay Area nonprofit serving refugee families, faced significant operational hurdles that impeded program scalability and support quality.

Information Barriers: Refugee families struggled to access real-time program information due to language barriers and fragmented communication channels.
Volunteer Knowledge Gaps: With 50+ rotating volunteers, ensuring consistent knowledge of operational procedures was nearly impossible — leading to recurring policy violations and session inconsistencies.
Administrative Burnout: Lead tutors spent 2+ hours every week manually reviewing Progress Logs and extracting insights by hand.

The Solution

I designed a Dual-Pillar AI Support Ecosystem connecting two integrated systems to transform SEED Academy's operations from manual and reactive to automated and proactive.

Real-Time Support (AI Chatbot) — A 24/7 multilingual AI assistant trained on SEED Academy's operational knowledge base, accessible via web-based QR code with zero app installation required.
Operational Intelligence (AI Agent) — An automated pipeline connecting Google Sheets, Zapier, and Gemini to autonomously analyze Progress Logs and deliver weekly student insight reports to the centralized Seed email.

① REAL-TIME SUPPORT

24/7 AI-powered chatbot providing instant operational guidance and overcoming language barriers.

② DATA INTELLIGENCE

End-to-end automation pipeline using Zapier and Gemini to transform raw logs into actionable insights.

My Process

I applied the SAM (Successive Approximation Model) to this project to ensure rapid deployment and continuous refinement of operational workflows. Each phase was built directly on field feedback to bridge the gap between complex operations and real-time user support.

Phase 1 — Preparation (Savvy Start)

Information Audit (Knowledge Base)

I conducted an Information Audit, identifying over 30 frequently recurring operational questions. This formed the foundation for the AI-optimized knowledge base used to train the chatbot, ensuring accurate and context-specific responses.

User Support Strategy (AI Chatbot)

Through Constraint Discovery, I identified that app installation was a significant access barrier for refugee families. This led to the adoption of a web-based QR code system, providing instant, zero-installation access for users.

Workflow Analysis (AI Agent)

I mapped the manual Progress Log workflow and confirmed that operational staff were spending 2+ hours every week on repetitive manual reviews. This analysis validated the critical need for an automated analytics pipeline.

AI Chatbot Access Flow

Phase 2— Iterative Design

AI System Prototyping (AI Chatbot)

Using the first version of the knowledge base, I built an initial chatbot prototype on Chatbase and measured answer accuracy across all 30 question categories. Edge cases and knowledge gaps were identified and continuously fed back into the knowledge base for refinement.

Automation Pipeline Engineering (AI Agent)

I designed and successfully tested an end-to-end automation pipeline: Google Sheets → Zapier → Gemini → Automated Email Delivery (every Saturday). This system was iteratively refined to ensure seamless data flow and accurate extraction of student insights.

Chatbot prototype

AI-Powered Progress Log Pipeline

Phase 3— Iterative Development

Role-Identification Logic (AI Chatbot)

Following beta testing feedback, I redesigned the chatbot's role-identification flow so that every conversation begins by identifying whether the user is a Refugee Family, Volunteer Tutor, or Operational Staff. This eliminated confusion and significantly improved response relevance by delivering tailored information from the very first message. approachable instructor presence that reduced cognitive load and increased overall engagement.

Advanced Prompt Engineering (AI Agent)

Leveraging Gemini, I iterated through multiple prompt versions to move beyond simple summarization, refining the logic to generate structured, actionable insights. I also engineered a delay & consolidation function to deliver a single, integrated email to the official Seed account every Saturday, replacing fragmented real-time notifications.

Chatbot's role-identification flow

A delay function

Experience the Project

Full development

FULL DEVELOPMENT

The following showcases the automated support infrastructure of the SEED Academy Smart Operations Ecosystem — engineered to streamline communications and operational workflows through real-time AI integration.

① 24/7 Operational Assistant (AI Chatbot)

This role-based AI Chatbot serves as the primary support line, eliminating the need for constant human intervention. It provides instant, multilingual guidance (in Dari and Spanish) on operational policies and safety protocols. By automating these recurring inquiries, operational staff save over 2 hours weekly, allowing them to focus on high-impact administrative tasks and program quality.

② Automated Insights Pipeline

An automated analytics pipeline was engineered using Google Sheets, Zapier, and Gemini, eliminating the labor-intensive process of manual log reviews. Every Saturday at 3:00 PM, the system autonomously analyzes all submitted progress logs and delivers a structured student insight report directly to the official Seed Academy email. By implementing a custom delay & consolidation function, I replaced noisy, fragmented notifications with a single, clean weekly summary, ensuring the leadership team can access centralized, data-driven insights in one place.

PAST: Manual Log Analysis

PRESENT: AI-Automated Reporting

Experience the Project

Results and Takeaways

From fragmented manual processes to a unified, AI-driven ecosystem. Below are the measurable results and key lessons learned from building a barrier-free support system.

1. Measurable Impact

From 2+ Hours to 0 Minutes: Completely automated the administrative workflow, eliminating over two hours of weekly manual log reviews. Reports are now instantly generated and delivered to the official SEED email, providing Lead Tutors with immediate access to student insights.

100% Barrier-Free Adoption: Successfully deployed the Dual-Pillar AI Infrastructure without requiring a single app installation. By utilizing simple web links and QR codes, the system achieved immediate adoption among users with varying tech literacy.

2. Human-Centric Outcomes

Bridging the Information Gap: Empowered non-English-speaking refugee parents with a 24/7 multilingual AI assistant, ensuring they always have access to vital program details like ride schedules and absence policies.

Actionable & Empathetic Insights: Transformed fragmented, manual progress logs into structured weekly briefings. By converting raw data into actionable insights, Lead Tutors can now provide deeper, more data-driven mentorship for students.

Empowering Volunteers: By removing the friction of complex administrative tasks, operational staff and volunteers can now focus 100% of their energy on what truly matters: connecting with and educating the students.

3. Key Takeaways & Lessons Learned

Technology Must Serve Accessibility: I learned that even the most advanced AI solutions fail if they don't account for the end-user's constraints. Designing for minimal friction—no logins, no app downloads—was the absolute key to successful adoption.
The Power of Agile Prototyping (SAM Model): Applying the Successive Approximation Model (SAM) taught me that rapid prototyping and continuous iteration are far more effective than aiming for initial perfection, especially in a dynamic nonprofit environment.
AI as a Strategic Partner: I discovered how to position AI not just as a simple summarization tool, but as a reliable, 24/7 operational partner that actively solves the chronic issue of understaffing in nonprofit organizations.

AI-Driven Operations: Automated Learning Support & Workflow Ecosystem

Audience: Operational Staff and Program Leadership at Seed Academy

Role: AI Systems Integration, Operational Workflow Automation, Learning Technology, Prompt Engineering, User Support Strategy

​Tools & Frameworks: Chatbase, Zapier, GPT-4o, Gemini, Camtasia, Canva, Claude AI, SAM (Successive Approximation Model)

The Problem

SEED Academy, a Bay Area nonprofit serving refugee families, faced significant operational hurdles that impeded program scalability and support quality.

Information Barriers: Refugee families struggled to access real-time program information due to language barriers and fragmented communication channels.

Volunteer Knowledge Gaps: With 50+ rotating volunteers, ensuring consistent knowledge of operational procedures was nearly impossible — leading to recurring policy violations and session inconsistencies.

Administrative Burnout: Lead tutors spent 2+ hours every week manually reviewing Progress Logs and extracting insights by hand.

The Solution

I designed a Dual-Pillar AI Support Ecosystem connecting two integrated systems to transform SEED Academy's operations from manual and reactive to automated and proactive.

​Real-Time Support (AI Chatbot) — A 24/7 multilingual AI assistant trained on SEED Academy's operational knowledge base, accessible via web-based QR code with zero app installation required.

​Operational Intelligence (AI Agent) — An automated pipeline connecting Google Sheets, Zapier, and Gemini to autonomously analyze Progress Logs and deliver weekly student insight reports to the centralized Seed email.

My Process

I applied the SAM (Successive Approximation Model) to this project to ensure rapid deployment and continuous refinement of operational workflows. Each phase was built directly on field feedback to bridge the gap between complex operations and real-time user support.

Phase 1 — Preparation (Savvy Start)​​

​

Information Audit (Knowledge Base)

I conducted an Information Audit, identifying over 30 frequently recurring operational questions. This formed the foundation for the AI-optimized knowledge base used to train the chatbot, ensuring accurate and context-specific responses.

​

User Support Strategy (AI Chatbot)

Through Constraint Discovery, I identified that app installation was a significant access barrier for refugee families. This led to the adoption of a web-based QR code system, providing instant, zero-installation access for users.

Workflow Analysis (AI Agent)

I mapped the manual Progress Log workflow and confirmed that operational staff were spending 2+ hours every week on repetitive manual reviews. This analysis validated the critical need for an automated analytics pipeline.

AI Chatbot Access Flow

Phase 2— Iterative Design

​​

AI System Prototyping (AI Chatbot)

Using the first version of the knowledge base, I built an initial chatbot prototype on Chatbase and measured answer accuracy across all 30 question categories. Edge cases and knowledge gaps were identified and continuously fed back into the knowledge base for refinement.​

​

Automation Pipeline Engineering (AI Agent)

I designed and successfully tested an end-to-end automation pipeline: Google Sheets → Zapier → Gemini → Automated Email Delivery (every Saturday). This system was iteratively refined to ensure seamless data flow and accurate extraction of student insights.

Chatbot prototype

AI-Powered Progress Log Pipeline

Phase 3— Iterative Development

​​

Role-Identification Logic (AI Chatbot)

​

Advanced Prompt Engineering (AI Agent)

Chatbot's role-identification flow

A delay function

FULL DEVELOPMENT

The following showcases the automated support infrastructure of the SEED Academy Smart Operations Ecosystem — engineered to streamline communications and operational workflows through real-time AI integration.

① 24/7 Operational Assistant (AI Chatbot)

② Automated Insights Pipeline

PAST: Manual Log Analysis

PRESENT: AI-Automated Reporting

Results and Takeaways

From fragmented manual processes to a unified, AI-driven ecosystem. Below are the measurable results and key lessons learned from building a barrier-free support system.

1. Measurable Impact

​

​​From 2+ Hours to 0 Minutes: Completely automated the administrative workflow, eliminating over two hours of weekly manual log reviews. Reports are now instantly generated and delivered to the official SEED email, providing Lead Tutors with immediate access to student insights.​

100% Barrier-Free Adoption: Successfully deployed the Dual-Pillar AI Infrastructure without requiring a single app installation. By utilizing simple web links and QR codes, the system achieved immediate adoption among users with varying tech literacy.​

2. Human-Centric Outcomes

​

Bridging the Information Gap: Empowered non-English-speaking refugee parents with a 24/7 multilingual AI assistant, ensuring they always have access to vital program details like ride schedules and absence policies.​​

Actionable & Empathetic Insights: Transformed fragmented, manual progress logs into structured weekly briefings. By converting raw data into actionable insights, Lead Tutors can now provide deeper, more data-driven mentorship for students.​​

Empowering Volunteers: By removing the friction of complex administrative tasks, operational staff and volunteers can now focus 100% of their energy on what truly matters: connecting with and educating the students.

3. Key Takeaways & Lessons Learned

​

Technology Must Serve Accessibility: I learned that even the most advanced AI solutions fail if they don't account for the end-user's constraints. Designing for minimal friction—no logins, no app downloads—was the absolute key to successful adoption.

The Power of Agile Prototyping (SAM Model): Applying the Successive Approximation Model (SAM) taught me that rapid prototyping and continuous iteration are far more effective than aiming for initial perfection, especially in a dynamic nonprofit environment.

AI as a Strategic Partner: I discovered how to position AI not just as a simple summarization tool, but as a reliable, 24/7 operational partner that actively solves the chronic issue of understaffing in nonprofit organizations.

AI-Driven Operations:
Automated Learning Support & Workflow Ecosystem

Tools & Frameworks: Chatbase, Zapier, GPT-4o, Gemini, Camtasia, Canva, Claude AI, SAM (Successive Approximation Model)

Real-Time Support (AI Chatbot) — A 24/7 multilingual AI assistant trained on SEED Academy's operational knowledge base, accessible via web-based QR code with zero app installation required.

Operational Intelligence (AI Agent) — An automated pipeline connecting Google Sheets, Zapier, and Gemini to autonomously analyze Progress Logs and deliver weekly student insight reports to the centralized Seed email.

Phase 1 — Preparation (Savvy Start)

Using the first version of the knowledge base, I built an initial chatbot prototype on Chatbase and measured answer accuracy across all 30 question categories. Edge cases and knowledge gaps were identified and continuously fed back into the knowledge base for refinement.

From 2+ Hours to 0 Minutes: Completely automated the administrative workflow, eliminating over two hours of weekly manual log reviews. Reports are now instantly generated and delivered to the official SEED email, providing Lead Tutors with immediate access to student insights.

100% Barrier-Free Adoption: Successfully deployed the Dual-Pillar AI Infrastructure without requiring a single app installation. By utilizing simple web links and QR codes, the system achieved immediate adoption among users with varying tech literacy.

Bridging the Information Gap: Empowered non-English-speaking refugee parents with a 24/7 multilingual AI assistant, ensuring they always have access to vital program details like ride schedules and absence policies.

Actionable & Empathetic Insights: Transformed fragmented, manual progress logs into structured weekly briefings. By converting raw data into actionable insights, Lead Tutors can now provide deeper, more data-driven mentorship for students.