From Doing the Work to Directing It: AI and the Assistant Editor with Kylee Pena

Summary

Adobe Live’s Kylee Pena and Ben Insler discuss how AI tools may change assistant-editor workflows and careers. Their central point is that AI can take on time-consuming tasks, but an editor or assistant still needs to define the work, check the results, and make creative and technical judgments.

Premiere AI Assistant: Current Capabilities

In the Premiere beta shown during the session, the AI Assistant can help with tasks such as:

• Renaming clips, including using visual content as the basis for labels.
• Bulk-transcribing interviews and adding markers—for example, at slates or when “action” is called.
• Organizing project media according to a requested structure.
• Syncing and stacking takes, though the speakers note that syncing still needs refinement and is not yet trained specifically for production workflows.
• Building a string-out or rough assembly for editors to review, rather than only sorting files into folders.

The assistant requires an internet connection. The speakers also note that some production environments are air-gapped and that the necessary models cannot yet run locally in those settings.

The assistant reports what it has done, and its changes are editable or undoable: generated sequences are ordinary sequences, and edits can be reversed. It may also flag files or situations it could not confidently handle rather than silently making every decision.

How to Direct the Assistant

Pena recommends treating the AI as a collaborator, not a vending machine or a conventional search engine. A useful brief should:

1. State the desired outcome.
2. Identify the relevant media and explain its creative intent.
3. Specify conventions such as naming, bins, and tracks.
4. Define the scope and ask the assistant to flag uncertainty.
5. Ask for a review of the result, then make the final call.

The speakers stress that interacting with an AI agent is often iterative: users can clarify, correct misunderstandings, and refine instructions. They suggest learning to define reusable “skills”—sets of instructions for recurring workflows—rather than viewing prompt engineering as a separate job.

Examples and Potential Uses

Insler describes possible uses for tasks that are more involved than a single command but still repetitive, including:

• Creating total-running-time (TRT) string-outs from multiple sequences while excluding leader or stray material.
• Comparing VFX shot IDs in an EDL or XML against a spreadsheet of approved shots.
• Helping parse and transform production documents or lists of tasks.

These examples illustrate how an AI agent could reduce repetitive setup and checking, while leaving the assistant editor responsible for defining the workflow and verifying the outcome.

Work, Skills, and Career Implications

The speakers acknowledge that automation could reduce some entry-level tasks, which have traditionally helped newcomers learn post-production. They argue that the deeper knowledge behind those tasks—why a workflow is structured a certain way, what footage matters, and what a particular editor or production needs—remains important.

They see AI as potentially giving teams more time for creative review, collaboration, and problem-solving rather than late-night technical chores. But they do not claim that jobs or working conditions are guaranteed to improve: market pressures may also lead employers to reduce staffing. They argue that assistant-editor roles are more likely to endure when their value is understood in terms of judgment, trust, communication, workflow knowledge, and accountability—not just button-pushing.

The speakers distinguish the ability to assemble footage from the editorial role itself: AI may be able to edit, but they believe human taste and storytelling judgment remain central. They also encourage industry professionals to advocate for sustainable roles and entry paths for the next generation.

Guide, Demo, and Product Notes

• Demo: Pena demonstrates the Premiere AI Assistant early in the session, showing marker creation, project organization, and batch clip renaming.
• Practical guidance: The discussion provides advice on writing briefs, supplying context, specifying conventions, and checking uncertain results.
• Availability: Pena says the Premiere AI Assistant is in public beta and can be tried by downloading the beta. She also mentions AI assistants in Photoshop, Illustrator, InDesign, and Frame.io, with work underway on one for After Effects.

The session is a discussion and demonstration, not a formal product review. The speakers emphasize that the technology is developing and may not handle every task reliably.

Main Speakers and Sources

• Kylee Pena — Adobe Premiere product marketing manager; former editor and workflow-technology professional.
• Ben Insler — Adobe Premiere team quality engineer; former assistant editor in film and television.