Meet the New Reelsy Video Agent: From One Prompt to a Finished Product Video

Most AI video tools begin with a form.

Choose a model. Pick a duration. Write a prompt. Configure an aspect ratio. Upload a reference. Submit the job. Wait. Then open another page to find the result.

That works when you already know exactly how the tool expects you to think. It is much less useful when you start with a normal creative request:

“Turn this product photo into a premium vertical ad.”

The new Reelsy Video Agent is designed around that request instead of the form. You describe the result, upload the material you already have, and continue in one persistent conversation while Reelsy AI plans the work, runs the production steps, and organizes the real outputs in the same workspace.

This guide explains what changed, how the two launch Skills work, what you will see during production, and how to get better results from your first prompt.

What Is the Reelsy Video Agent?

The Reelsy Video Agent is a conversational workspace for producing short product videos.

It combines four parts that are usually split across different screens:

1. Conversation — the brief, clarification questions, revisions, and status explanations
2. Skills — focused production methods for a specific type of video
3. Tasks — the real production steps currently running or waiting
4. Outputs — generated images, supporting artifacts, and the final video

The conversation is the control surface. The outputs are the result. You do not need to translate your idea into internal workflow settings before you begin.

At launch, the Video Agent is optimized for 8-second, 9:16 vertical videos and includes two production-ready Skills:

• Product ad
• UGC video

Both start from natural language and can use an uploaded image, but they solve different creative jobs.

What Changed in the New Experience

The previous video-agent experience exposed too much implementation detail. Planning facts, provider jobs, intermediate artifacts, and final results competed for attention as similar-looking cards. A creator could see that the system was busy without always understanding what mattered next.

The new experience introduces a clearer product model.

Chat Comes First

The left side of the experience is a real continuing conversation, not a one-time prompt box. Reelsy AI can ask for missing information, explain what it is doing, report a recoverable failure, and respond to a revision request in the same session.

You can begin with a short request. If the brief is already sufficient, production can start without forcing an unnecessary confirmation step. If a critical detail is missing, the Agent asks a small number of focused questions instead of presenting a large configuration form.

The Workspace Shows Real Production State

The production workspace no longer invents placeholder outputs to make the screen look busy. It projects real saved results and real task status.

That means a generated image appears only after an image result exists. A final video appears only after the video has been produced and saved. Progress information stays distinct from output.

Sessions Are Persistent

Every meaningful conversation can be reopened from Recent sessions. Titles are derived from the first real request, so the history list contains names such as:

• “Create a premium coffee cup product ad”
• “Make a casual skincare UGC video”
• “Turn this shoe image into a vertical launch video”

Empty connections are not treated as conversations, and a refresh does not require you to reconstruct the brief from memory.

Outputs Use a General Artifact Model

The workspace is not hard-coded to one video recipe. Images, videos, text outputs, previews, and future Skill results can use the same artifact boundary while choosing the right renderer for their media type.

For creators, the practical effect is simple: the workspace can grow beyond today’s two Skills without turning every new capability into a completely separate product UI.

Product Ad vs. UGC Video

Choosing the right Skill gives the Agent a better production objective before it writes the plan.

Use Product Ad When the Product Is the Hero

Product ad works best when the object, packaging, material, or feature needs to stay visually central.

Good requests include:

• “Create a premium launch ad for this watch. Use dark studio lighting and a clean metallic look.”
• “Turn this skincare bottle into a hook-first product demo for TikTok.”
• “Make a minimal product video that opens with the packaging and ends on the texture.”

Include the positioning you actually know. If a claim is not visible in the source image and you have not supplied it, do not rely on the Agent to invent it. Prices, certifications, review counts, discounts, and performance claims should come from your brief or source material.

Use UGC Video When the Creator Perspective Is the Hero

UGC video works best when the ad should feel like a person discovered, used, or explained the product.

Good requests include:

• “Create a casual UGC demo for these earbuds. Open with a frustration about calls in noisy places.”
• “Make a natural skincare testimonial with a close-up product demonstration.”
• “Create a hands-on UGC video for this kitchen tool. Keep the tone practical, not polished.”

The most useful UGC briefs name the audience, the situation, and the delivery style. “Busy parents,” “first-time founders,” and “people working in shared offices” are more actionable than “make it viral.”

How to Create Your First Video

Open the Reelsy Video Agent from the dashboard.

Step 1: Describe the Result

Start with the outcome you want. A strong first prompt usually contains:

• The video type
• The product or subject
• The audience or use case
• The opening hook
• The desired visual tone
• Any non-negotiable product facts

For example:

“Create an 8-second vertical product ad for this insulated coffee cup. Target commuters. Open on the leak-proof lid, use bright morning light, and finish with the cup in a train-station setting.”

You do not need to describe every camera move. The Agent should own the production plan. Your job is to make the creative intent and factual constraints clear.

Step 2: Upload a Useful Reference

Upload a product or creator image when visual consistency matters.

A useful source image has:

• A clear main subject
• Enough resolution to see shape and material
• Limited obstruction
• Lighting that does not hide important details
• A composition that can be adapted to vertical video

The image is evidence, not decoration. Reelsy AI analyzes visual details before it writes a media plan when those details matter to the request.

Step 3: Choose a Video Type

Choose Product ad or UGC video from the homepage. Each option reveals practical starting directions that can prefill the prompt without sending it automatically.

You remain in control of the wording. Edit the prefilled request, add product facts, attach a source image, and submit when the brief is ready.

Step 4: Answer Only the Necessary Questions

If the Agent needs a critical detail, it asks a concise question in the conversation.

Examples:

• Which audience should the video target?
• Should the UGC delivery feel casual or expert-led?
• Which product benefit is safe to emphasize?

If the source and prompt already answer those questions, the Agent can move directly into production.

Step 5: Follow the Production Progress

Once production starts, the workspace shows the current task state. Depending on the Skill, the plan may prepare visual anchors before generating the video result.

You may see states such as:

• Planning the production
• Preparing a visual output
• Waiting for generation
• Recovering the latest saved state
• Result ready
• Needs attention

These states describe the workflow without exposing raw provider payloads or internal task identifiers as if they were creative output.

Step 6: Open the Final Output

When the final video is saved, it appears as a real output in the workspace. Open it to review the result.

The conversation remains available beside the result, so the next action can be expressed naturally:

• “Keep the product framing but make the opening more energetic.”
• “Use a warmer creator tone.”
• “The packaging color is wrong. Match the uploaded reference.”
• “Make the payoff clearer in the final beat.”

The existing session provides context for the revision. You do not need to start over simply because the first result needs adjustment.

Understanding the Workspace

The new workspace intentionally separates three concepts.

Conversation

Conversation contains what you and Reelsy AI said. It is where you clarify intent, receive explanations, and request changes.

Progress

Progress describes work that is currently happening. It can change as tasks move from planning to generation to completion.

Outputs

Outputs are saved artifacts. They can include supporting images and the final video. An output remains part of the session after a task finishes.

This separation prevents a common problem in agent interfaces: treating every event as a message and every internal record as a card. Creators should be able to identify the result without reading an execution log.

Returning to a Previous Session

Use the Recent list to reopen a conversation.

The list is ordered by real conversation activity and uses a title based on the first request. When a session opens, Reelsy restores the durable conversation and the saved production facts used by the workspace.

If a live response is interrupted by a temporary connection problem, refreshing the session restores the durable history. The interface does not automatically resubmit a generation request during recovery, which helps avoid duplicated work.

Prompting Patterns That Work

The best prompts are specific about intent and restrained about implementation.

Good Product Ad Pattern

Create a [tone] product ad for [product]. Target [audience]. Open with [visual hook]. Emphasize [verified benefit]. Use [visual direction]. End with [payoff].

Example:

“Create a clean, premium product ad for this travel bottle. Target frequent flyers. Open with a close-up of the locking cap, emphasize the slim carry-on shape, use cool airport lighting, and end with the bottle beside a cabin bag.”

Good UGC Pattern

Create a [delivery style] UGC video for [product]. The creator is speaking to [audience] in [situation]. Open with [problem or reaction]. Demonstrate [action]. End with [payoff].

Example:

“Create a casual UGC video for these wireless earbuds. The creator is speaking to remote workers in a shared apartment. Open with frustration about background noise, demonstrate putting in the earbuds before a call, and end with a calm reaction.”

Avoid Empty Adjectives

Words such as “cinematic,” “viral,” and “premium” can help, but they are not a complete brief.

Instead of:

“Make this viral and cinematic.”

Use:

“Open with a fast macro reveal, keep the background dark and minimal, use one clear product action, and finish on a bright hero shot.”

Concrete creative constraints give the Agent something it can plan.

Current Boundaries

The first release is intentionally focused.

• The primary workflow targets 8-second vertical videos.
• The workspace is optimized for desktop production.
• Product ad and UGC video are the two launch Skills.
• Uploaded images can ground visual planning; do not assume unsupported facts from an image.
• A completed output must come from a real generation result. The workspace does not fabricate preview media.
• Some requests may stop with a clear error if image analysis or an external generation step fails. Correcting the source or retrying a failed step is safer than silently producing an ungrounded result.

These boundaries make the workflow more predictable. The goal is not to expose every possible model option on day one. The goal is to make a small number of valuable video jobs work coherently from brief to result.

Frequently Asked Questions

Do I need to write a detailed video-generation prompt?

No. Describe the creative result, audience, hook, and verified product facts. Reelsy AI creates the production plan. Add camera or lighting detail only when it matters to the outcome.

Can I start with only an image?

You can upload an image, but a short sentence about the desired result will produce a clearer direction. Tell the Agent whether you want a Product ad or UGC video and what the opening should communicate.

Why did the Agent ask a question instead of generating immediately?

The request may be missing a decision that materially changes the result, such as audience, tone, or the benefit to emphasize. The Agent asks focused questions instead of guessing an important constraint.

Will refreshing duplicate my video generation?

The recovery path restores saved history and does not automatically resend the original generation request. This is designed to prevent duplicate execution.

Can I reopen a finished video later?

Yes. Reopen the conversation from Recent sessions. Saved outputs remain associated with that session.

Is this a general-purpose AI assistant?

It is a video-production Agent. Conversation is the interface, but the launch Skills are focused on Product ads and UGC videos rather than unrelated assistant tasks.

Start a Video Production Conversation

The central change is not a new prompt field or another model selector. It is continuity.

Your brief, questions, production progress, generated media, final result, and revision request now belong to the same persistent workspace.

Open the Reelsy Video Agent and start with the result you want to create.

If you do not have an account yet, create one here.