A 3-Step Workflow to Supercharge AI for Work and Side Projects: Gather, Speak, and Delegate

Many professionals and side-project creators invest in new AI models or paid prompt packages only to find that the generated outputs fall far short of expectations when applied to real tasks. On October 16, 2026, X creator Cheongchun (@dudrms606) shared an insightful summary based on YouTube channel Oppadu Excel's year-long journey of integrating AI into day-to-day operations, outlining a practical 3-step workflow—Gather, Speak, and Delegate—that focuses on establishing context before searching for complex prompts.

Image source: X @dudrms606 (출처: 유튜브 오빠두엑셀)

While managers routinely provide new hires with onboarding documentation, historical project files, and detailed standard operating procedures, they often treat AI very differently—expecting high-quality results from a single vague instruction like "do this well." However, getting exceptional results from AI depends far less on memorizing prompt hacks than on providing the model with the exact background context and operational logic needed to understand the task.

Step 1: 'Gather' — Consolidate Scattered Materials and Reference Samples

The initial step focuses on assembling the foundational materials and past deliverables required for the assignment.

• Centralizing Fragmented Assets: Gather reference documents, spreadsheets, previous work samples, and frequently used templates scattered across multiple folders into a single workspace.
• Establishing Quality Baselines: Rather than letting the model guess the expected format, provide concrete examples of completed, successful deliverables. Just as an onboarding employee needs reference examples of past reports, AI requires clear benchmarks to align with desired standards.

Step 2: 'Speak' — Record Workflows on Screen to Generate Operating Manuals

The second step is highlighted as the most practical and accessible technique in the entire workflow.

Writing a standard operating procedure (SOP) from scratch is often intimidating and time-consuming. In contrast, simply recording your screen while performing routine tasks—explaining out loud what steps you take, what details you skip, and why you make specific decisions—requires minimal extra friction because it follows your existing daily routine.

• Screen Capture with Spoken Commentary: Capture your live workflow along with spoken explanations to articulate the tacit knowledge and edge-case rules that usually exist only in your head.
• AI-Powered Manual Drafting: Feed the recorded transcript or spoken walkthrough into the AI and instruct it to synthesize the steps into a structured, step-by-step operating manual.
• Privacy and Data Hygiene: Before sharing screen recordings or captures with any AI model, verify that sensitive personal data, internal credentials, or proprietary client information are not visible on screen.

Step 3: 'Delegate' — Separate Initial Drafting from Final Human Verification

The final step establishes clear boundaries between what the AI handles and where human oversight is mandatory.

• Delegating Initial Drafts: Assign the AI to manage repetitive, time-intensive tasks such as consolidating raw data, formatting templates, and producing first drafts.
• Retaining Human Sign-off: Keep the final review, approval, and publication strictly in human hands.
• Prioritizing Accuracy Over Speed: For instance, while an AI assistant can effectively draft customer service or comment responses, the creator emphasizes reviewing and publishing each reply manually. Accurate, dependable communication holds far more long-term value than raw turnaround speed.

Ultimately, mastering AI tools begins not with chasing expensive prompt libraries, but with clearly articulating and structuring how you actually do your own work.

Original source

• Cheongchun (@dudrms606) on X: October 16, 2026 3-Step AI Workflow for Work and Side Projects (Referencing: Oppadu Excel on YouTube)