DAY 1 - Pelatihan Terampil Bekerja dengan AI: Praktik Mengelola Dokumen, Data, dan Presentasi

Summary

This first-day training session explains how AI tools can assist with preparing and managing workplace documents. The central message is to treat AI as a smart assistant that can speed up drafting, not as a replacement for human judgment. Users remain responsible for checking accuracy, suitability, and compliance before using or sending AI-generated work.

Main Ideas

• AI can help with many document tasks, including activity reports, proposals, official letters and memos, meeting minutes, teaching materials, summaries, and follow-up emails.
• Good results depend on useful context and clear instructions. Vague requests may lead to invented or incomplete details. Providing source documents, examples, templates, standard operating procedures (SOPs), notes, or spreadsheets helps AI produce more relevant drafts.
• AI can work with more than text. The presenter described using images of handwritten notes, screenshots of chats, photos of physical documents, PDFs, and Excel files as input.
• AI-generated content is only a draft. Users should verify facts, dates, names, figures, legal references, formatting, and language before relying on the result.
• Confidentiality matters. Participants were advised to follow their organization’s rules, avoid uploading sensitive information without authorization, and consider disguising identifying details or using aggregate data where appropriate. The presenter also mentioned privacy settings but cautioned users to follow their workplace policies.
• Claude and ChatGPT were demonstrated as commonly used tools. The presenter suggested choosing based on the task and the output style that works best for the user; neither was presented as universally superior.
• AI tools have context limits. For large or complex work, it is often better to divide the task into stages rather than expect the AI to handle everything in one long exchange.

Suggested Workflow for Prompting AI

The presenter’s framework can be organized around these elements:

1. Assign a role. State what perspective the AI should take, such as that of an administrative staff member.
2. Specify the document type. Say exactly what you want produced—for example, an activity report, meeting minutes, or official letter.
3. Provide information and source material. Attach relevant notes, templates, previous reports, SOPs, terms of reference, spreadsheets, or other references. Explain what each attachment contains and which one should guide the format or facts.
4. State the audience and purpose. Identify who will receive or use the document and why it is being prepared—for example, an accountability report for a department head.
5. Set constraints and rules. Specify requirements such as formal language, a page limit, required sections, table or bullet formatting, and instructions not to add information absent from the sources.

Practical Techniques Demonstrated

• Ask the AI to ask clarifying questions first if important details are missing, rather than letting it guess.
• Provide an existing template or prior document when a specific format must be followed.
• Work in stages: organize notes or data, create a draft, request revisions, and then combine or format the results.
• Use project or folder features, where available, to keep recurring reference materials and instructions together. This can reduce the need to re-enter the same context each time.
• Split larger jobs into manageable parts because the AI may lose earlier context in a long conversation.
• Request revisions—for example, to change a section into bullet points, adjust the introduction, or prepare a Word document.
• After preparing a report, ask the AI to draft a submission email or covering letter, but review it before sending.

Demonstration

The presenter showed how to create an activity report using handwritten notes, a prior report as a format reference, a terms of reference document, and an Excel participant recap. The tools produced drafts combining information from those sources. The presenter compared the results, noted that formatting and details still required checking, and demonstrated requesting further revisions and related correspondence.

Reviewing AI Output

Participants were urged not to trust AI output automatically. The discussion described three possible levels of review:

• Manual review: Read the output and compare it with the source material.
• AI-assisted review: Ask the AI to compare its draft against the original documents or a checklist.
• Rule-based or test-case review: Define specific criteria that the output must satisfy, then have the AI check and revise its work against them.

The presenter and moderator referred to a “5B” check before sending documents, but the subtitles do not clearly explain what the five items are.

Questions and Further Points

• For trial minutes, the presenter said AI could help if given the relevant records and instructions. Long case materials can be handled in stages or organized by case in a folder.
• For research proposals and academic writing, AI should be treated as a drafting aid. Whether paraphrasing or plagiarism checks are required depends on the relevant academic rules and review process.
• ChatGPT and Claude may both be suitable for thesis or proposal work; users should compare them against their needs and preferred output.
• AI can assist with web or mobile application development. An Excel-based process may also be translated into a web application, though the steps depend on the project.
• Short prompts can work when the necessary context is already available; the prompt itself does not replace missing information.
• Email connections may be available directly through AI tools. Building a custom system that accesses email or other services may require using an API.

Closing Message

The presenter characterized AI as a multiplier: it can amplify a user’s existing knowledge and capabilities, but it can also amplify mistakes when the user lacks the expertise to assess its output. Participants were encouraged to experiment with AI while maintaining responsibility for the results.

Speakers and Sources Featured

• Ana — event host and moderator.
• E. Siswanti — education program officer who gave the opening remarks.
• Bima Putra Pratama — presenter and AI practitioner leading the training and demonstration.
• Rah Nur Samaani (name rendered inconsistently in the subtitles) — participant who asked about preparing trial minutes.
• Ricky — participant who asked how to evaluate AI-generated work.
• Nurdiansari — participant who asked about AI use in research proposals and paraphrasing.
• Sugeng Wahono — participant who asked about creating web applications with AI.
• Roroya — participant who asked which tool to use for thesis writing.
• Theopilus Ardian — participant who asked about converting Excel-based asset reports into web applications.
• Rita Meliani — participant who asked how to write concise prompts while reducing hallucinations.
• Muhammad Akbar — participant who asked about connecting AI to email and API requirements.
• Edukasa promotional song/anthem — featured in the opening and closing music.