100. That's how many "director-grade expert skills" are available on demand in a newly launched professional video agent platform.

On July 14, what's being called the "world's largest professional video agent Skill Hub" officially launched. This isn't another "text-to-video" tool — at its core is a collaborative canvas workflow: you drag and drop clips, arrange transitions, add subtitles, and adjust pacing on a canvas, while 100+ AI agent "specialists" stand by — storyboard experts, color graders, pacing editors, music supervisors, subtitle designers.

You don't need to know how to edit video. You just direct like a filmmaker: "This section is too slow, tighten it up" "Add a suspenseful transition here" "Give the whole thing a cinematic color grade" "Find me a light, upbeat BGM track." Agents understand your intent, execute automatically, and you review and tweak.

Three Stages of AI Video: Toy → Tool → Workbench

AI video has evolved through three distinct phases to reach today.

Phase 1: Generation toys (2023-2024). Represented by early Sora demos, Runway Gen-2, Pika. The defining trait was "impressive but useless" — input text, get a few seconds of video, looks cool, but too short, too inconsistent, too uncontrollable to use in finished productions. Mostly used for novelty short-form social content.

Phase 2: Clip tools (2025-early 2026). Represented by Kling, Jimeng, and the official Sora release. Traits: "usable but siloed" — you could generate minutes of video, character consistency and physics improved, but AI generation was just one step in the production pipeline. You still downloaded clips, imported into Premiere or CapCut, manually edited, color graded, scored, subtitled. AI made the "ingredients" but you still had to "cook."

Phase 3: Workflow platforms (starting mid-2026). This is what the new video agent represents — AI is no longer just a "clip generation tool." It's taking over the entire creative workflow. From script to storyboard, clip generation to editing, color grading to scoring to subtitles, AI agents participate throughout. You're the director; they're your film crew.

Why Workflow Matters More Than Generation Quality

Many people still compare "who generates sharper video" or "who has better consistency," but professional creators know video production is never just about "generating a clip."

A three-minute short video involves dozens of decisions: how does the opening hook attention? Where do you pause? Where do transitions go? How do you control pacing? Where does BGM enter and exit? What subtitle font? How do you add text overlays without being jarring? How do you unify the color grade? These decisions together make the final video quality.

Previous AI video tools solved "do we have clips" — now video agents solve "how do we turn clips into good content". That's a qualitative leap.

Another notable signal at the same time: Bilibili's AI Creation Open Competition officially launched its "coin leaderboard," with monthly participants surpassing 5,000. This shows AI video creation has moved from niche geek toy to something masses of regular users are trying. When AI lowers video creation's barrier from "three years to learn editing" to "if you can talk, you can make it," the supply side of content production is heading for an explosion.

Of course, today's video agents aren't perfect — complex narrative, fine-grained control, long-form logical consistency still need work. But the direction is clear: AI video is no longer about "generating something fun" — it's genuinely becoming a productivity tool for content creation.

Directors, your AI film crew is ready.

See you tomorrow.