Okay team, let's dive into something we're all navigating:
Making AI actually work for us in video production, not just a waste of time.
Beyond the Hype: Making AI a Real Co-Pilot in Your Video Workflow
We're all seeing AI tools flood our NLEs, scriptwriting software, and project management platforms. Maybe you've tried using ChatGPT for script outlines and gotten generic fluff, or asked an AI editor for selects only to spend more time correcting its choices than making them yourself. Is AI feeling more like a time-sink than the biggest game-changer since the Canon 5D MarkII??? (If you know, you know)
I've definitely been there. I've spent too much time wrestling with these tools, trying to bend them to fit our creative standards and client standards. The potential is obvious, but the path often feels frustratingly unclear.
What made it all 'click' for me? Getting serious about prompt engineering.
Prompt engineering or "Prompting", is the crucial skill for effectively communicating our vision and requirements to these powerful new collaborators. It’s how we move from useless outputs to genuinely useful results - or at least getting us to the refining stages quicker saving immense amount of time. In this post, I want to share what I've learned – a practical breakdown of how to craft prompts that get AI working for you, specifically in our demanding video workflows.
(And hey, if you find this useful, I’m digging into more advanced techniques and tool breakdowns in my regular newsletter – a space for us pros to share what's actually working. More on that later.)
Quick Refresher:
Why Understanding the Engine Matters for Steering
Before the 'how,' a quick 'what'. Most tools we're using (ChatGPT, Gemini, Claude, etc.) are powered by models like GPT (Generative Pre-trained Transformer). Think of them as incredibly sophisticated text prediction engines trained on vast datasets. They excel at understanding and generating human-like language.
Why does this matter for us visual storytellers? Because even when we want visual outputs (shot lists, edit notes, color palettes), we're instructing the AI through text. Effective prompting means translating our visual and logistical needs into clear, structured language this engine can accurately interpret.
Why Prompt Engineering is Non-Negotiable for Video Pros
AI is embedding itself across our entire pipeline. Getting prompts right unlocks tangible benefits:
- Reclaim Creative Time: Automate the tedious stuff – transcript analysis, initial scheduling drafts, metadata tagging – freeing up headspace for craft and storytelling.
- Spark & Refine Ideas: Use AI as a tireless brainstorming partner for concepts, visual metaphors, B-roll ideas, or even initial music briefs.
- Streamline Operations: Generate clearer communication documents (call sheets, client updates, approval trackers) faster and more consistently.
But these advantages only materialize when we direct the AI effectively. That's prompt engineering: moving beyond simple requests to precise, context-rich instructions.
The 3 Elements of a High-Impact Prompt
Getting consistent, high-quality results boils down to consciously structuring your prompts around three core elements:
1. Context: Setting the StageThis is where you brief the AI, giving it the necessary background and defining its role. Think: If I hired an expert for this, what would they absolutely need to know first?
Key Components:
- Role: Define the AI's persona
- "You are an experienced documentary film editor"
- "You are a commercial producer managing client relations"
- Background: Provide essential project details
- Client, goals, target audience, key constraints, existing materials, etc.
- Format/Constraints:
- Specify limitations - Desired length, tone, style, required inclusions/exclusions, etc.
Example: "You are a post-production supervisor finalizing deliverables for a corporate brand film. The client requires broadcast-safe levels and specific caption formatting (CEA-708). The project was shot on ARRI Alexa Mini LF, edited in DaVinci Resolve."
2. Task: Defining the ActionThis is the clear, specific instruction about what you need the AI to do. Ambiguity is your enemy here.
Key Components:
- Action Verb: Use precise verbs
- "Analyze," "Generate," "Summarize," "Compare," "Create," "List," "Outline"
- Specific Requirements: Detail the exact criteria the AI must meet.
- Steps (if complex): Break down multi-part tasks logically.
- Desired Outcome: Briefly describe what success looks like.
Example (Continuing from above): "Analyze the attached technical delivery specs document. Create a checklist of the essential export settings required in DaVinci Resolve to meet these broadcast and captioning requirements for the final ProRes 4444 master file."
3. Output: Specifying the DeliverableThis tells the AI how you want the information presented. Don't leave it to chance; structure leads to usability.
Key Components:
- Format: Define the structure
- Bulleted list, table, numbered steps, JSON, script format, paragraph
- Level of Detail: Specify desired length or granularity
- Concise summary, detailed analysis, word count
- Tone/Style: Indicate the required voice
- Professional, technical, concise, encouraging
Example (Completing the prompt): "Present this checklist as a table with columns for 'Setting Category' (e.g., Video, Audio, Captions), 'Parameter,' 'Required Value,' and 'Notes/Rationale.' Ensure the language is clear and technical."
Complete CTO Prompt:
"You are a post-production supervisor finalizing deliverables for a corporate brand film. The client requires broadcast-safe levels and specific caption formatting (CEA-708). The project was shot on ARRI Alexa Mini LF, edited in DaVinci Resolve. Analyze the attached technical delivery specs document. Create a checklist of the essential export settings required in DaVinci Resolve to meet these broadcast and captioning requirements for the final ProRes 4444 master file. Present this checklist as a table with columns for 'Setting Category' (e.g., Video, Audio, Captions), 'Parameter,' 'Required Value,' and 'Notes/Rationale.' Ensure the language is clear and technical.
Moving Beyond Simple Tasks: Leveraging Project Context
The real power emerges when you provide the AI with rich, project-specific context before asking it to perform tasks. Think about feeding it meeting notes, past client emails, approved scripts, transcripts, or even style guides.
Example Integration: Use the 'Resource Management' prompt from the original post ("Before I prompt you... respond with ‘copy that’...") to load multiple documents, then ask the AI to synthesize information across them (e.g., "Based on the client emails and the approved script provided, generate a list of key B-roll shots needed, categorized by scene").
Want a shortcut? Mastering these elements takes practice. To help, I'm put together a Prompt Engineering Cheat Sheet for Video Pros with 10+ templates for common tasks like interview analysis, shot listing, and client comms. Signup below to get an email with instant access as soon as it's ready.
Prompt Engineering in Action: Examples from the Field
Let's make this concrete with examples we can all relate to:
For Editors:
Problem: Quickly finding the narrative gold in hours of interview footage.
- Prompt: "You are a seasoned documentary editor. Analyze the attached interview transcript [Paste Transcript or reference uploaded file]. Identify the 5 main narrative themes discussed. For each theme, provide 2-3 key soundbites (include timecode, speaker, and verbatim quote) that best represent it. Present the output as a table with columns: 'Theme,' 'Timecode,' 'Speaker,' 'Key Quote.'"
- Result: A structured starting point for your edit, saving hours of initial logging and highlighting core story beats.
For Producers:
Problem: Ensuring technical settings are correct for specific camera/lens/delivery combinations.
- Prompt: "You are a post-production tech expert. My project was shot on a Sony FX6 using Sirui 2.39:1 anamorphic lenses. The final delivery needs to be a 1920x1080 ProRes 422 HQ file with the correct desqueeze applied for web viewing. List the critical project settings, timeline settings, and export settings required in DaVinci Resolve 18.6. Present as a checklist with 'Setting,' 'Value,' and 'Brief Rationale.'"
- Result: Precise technical guidance tailored to your specific workflow, minimizing errors and ensuring quality.
For Producers/Project Managers:
Problem: Streamlining the often tedious client feedback and approval process.
- Prompt: "You are a client-facing video project manager. Based on the attached final script and storyboards [Reference uploaded files], create a client approval checklist in a table format. Include columns for 'Scene/Element,' 'Description,' 'Client Action Required' (e.g., Approve Concept, Provide Feedback on Graphics), 'Status' (Pending, Approved, Revisions Needed), and 'Client Notes.' Add a brief introductory sentence explaining how the client should use this document."
- Result: A clear, professional document that simplifies client communication and tracks approvals efficiently.
Level Up Your Workflow: Prompting is Our New Essential Skill
Let's be real: AI isn't going anywhere. Mastering prompt engineering moves us from being reactive users to proactive directors of these powerful tools. It's how we ensure AI enhances our creativity and efficiency, rather than dictating or hindering them. It’s about working smarter, reclaiming time for the craft we love, and ultimately, delivering better work.
This field is evolving incredibly fast. New tools, techniques, and ethical considerations pop up constantly. If you want to stay ahead of the curve and explore how we, as video professionals, can best leverage these advancements...
Join the conversation in the Coffee On Set Newsletter. Every other week, I share techniques from the field, AI strategies, and more insights that you can quickly go through as you sip on your morning brew ☕️