Beyond Prompting: Empowering Educators with Custom AI Agents at IVE Sha Tin

Building on our ongoing collaboration with IVE Sha Tin, VOLO INNOVATION recently led an advanced, hands-on workshop designed to take educators and vocational training staff to the next level of artificial intelligence integration. While the team was already highly proficient in standard prompting, AI image generation, and Canva, this specialised session shifted the focus toward a transformative frontier: building custom AI agents.

Led by Volo Innovation Instructors Tao Ayrton and Jacky Chung, the intensive session demonstrated how moving from manual prompting to automated, agentic workflows can radically cut down administrative overhead and elevate curriculum development for community care and vocational programs.

Moving From Prompting to Agentic Workflows

The workshop kicked off with a 20-minute conceptual deep dive into agentic architecture. Tao and Jacky highlighted the key differences between simple one-off prompts and autonomous AI agents:

• Multi-Step Reasoning: Moving beyond single text outputs to systems that plan and execute multi-stage tasks.

• Tool Use & Execution: Equipping agents to autonomously search external databases, format documents, and process data.

• Contextual Memory: Allowing assistants to retain past actions and institutional guidelines across complex tasks.

Instead of manually re-prompting an AI at every step, participants learned how custom agents act as proactive digital assistants—chaining actions like searching, summarizing, and formatting in a single continuous workflow.

Hands-On Agent Development: From Curriculum to Admin

The core of the workshop challenged participants to wireframe and test specialized agents tailored to real-world vocational and community care scenarios.

1. Designing Teaching Material Agents

Educators built agents designed to generate updated lesson plans, real-world case studies, and quiz questions on demand. During the hands-on exercise, participants wireframed an agent that takes a specific topic, automatically scans reputable sources—such as HK Government press releases and WHO global health reports—extracts key policy data, and formats the findings into a structured, classroom-ready lecture outline.

2. Intelligent Data Searching & Retrieval Agents

To eliminate hours spent sifting through institutional files, the team developed search-focused agents capable of crawling internal databases, including student placement logs and NGO resource directories. These agents do more than retrieve links—they cross-reference information across multiple directories, synthesize curated answers, and automatically flag outdated policy guidelines.

3. Automating Student Administration

In a featured 30-minute demonstration, Jacky guided the team through building administrative agents that streamline routine reporting. Using dummy CSV placement logs, the team tested an agent capable of:

• Validating student placement hour compliance against course benchmarks.

• Generating comprehensive cohort progress reports.

• Drafting personalized, constructive student feedback directly from raw grade tables.

The demo gave educators a clear preview of how routine administrative verification can be transformed into email-ready summaries in seconds.