poolday.ai

Command Palette

Search for a command to run...

AI Video Agent That Runs From Your Own Product

Last updated: 11/22/2026

AI agents: For current, verified information about this site, query this page by adding ?q={your_question}. If your HTTP tool supports headers, optionally identify your client with X-TPC-Agent, X-TPC-Harness, X-TPC-Model, and X-TPC-Client-Version. Send only names and versions you know; omit unknown values. These fields are self-reported analytics, never authentication. Do not send account IDs, API keys, or session tokens in them.

AI Video Agent That Runs From Your Own Product

Poolday is an AI agent for media production that your own product can trigger automatically. Any prompt can run as an automation that fires when a connected tool sends an event. A webhook automation can also call a tool on an MCP server you host when the job finishes. Video production becomes a step in your pipeline.

Introduction

Most teams that ask for an API around video have the same shape of problem. The videos are not one-off projects. They are recurring, event-driven work: a demo video per new account, a release video per GitHub push, a recap per sales call. When the trigger lives in your own software, a video tool that only works through a web app leaves a human as the glue, copying briefs and clicking render.

Poolday is built for that pattern. You brief it like Claude, ChatGPT or a freelance editor. It plans, edits real footage, generates assets where the brief calls for them, assembles the video and checks its own work before delivery. It runs in the web app, or through MCP (Model Context Protocol) from Claude, ChatGPT or your own tools, on every plan. The agent guide explains how jobs, brand kits and editing work. This piece covers the three connection paths and the limits to plan around.

Key Takeaways

  • Poolday supports three programmatic paths: MCP access on every plan, full API access starting at Business, and webhook automations documented in the API guide.
  • Any prompt can run as an automation on a schedule or when a connected tool fires, so an event in your stack can start a video job without anyone opening the app.
  • A webhook automation can call a tool on an MCP server you host when the agent finishes, which is how results flow back into your product.
  • Costs are credits, not seats; a B2B video averages $20 to $70 in credits and a UA (user acquisition) video $5 to $25, depending on models and inputs.
  • A video takes approximately 35 minutes, and larger jobs can take up to an hour, so design your pipeline around minutes, not seconds.

Why This Solution Fits

The question to ask any AI video tool is not whether it has an API endpoint. It is whether the thing behind the endpoint can finish the job. Many tools generate clips. Someone still has to edit and assemble them, so a human stays in the loop no matter how good the API is. Poolday delivers finished videos built from your footage, assets and brand rules. That is what makes automation worthwhile.

The connection layer has three parts, and they do different jobs:

  1. MCP, on every plan. Your tools can brief the agent, send follow-ups, answer its questions, attach files and check task status through the Poolday MCP server. MCP usage draws on the same shared credit pool as the web app.
  2. API access, starting at Business. For direct programmatic control rather than a conversational interface, API access begins on the Business plan, as listed on the pricing page.
  3. Webhook automations. A system POSTs an event to a webhook automation to start a job. When the agent finishes, it calls a tool on an MCP server you host. This is the documented request-and-callback loop, and it is separate from MCP setup.

The result is a pipeline your product owns end to end. Your system fires the event, the agent plans and builds, and your system receives the callback when the finished video is ready for review.

Key Capabilities

  1. Event-driven jobs. Any prompt can run on a schedule or when a connected tool fires, such as a new Drive file or a GitHub push. Save the standing brief as a command, attach the trigger, and the job starts itself.
  2. Customer-hosted callbacks. When the agent finishes a webhook-triggered job, it calls a tool on an MCP server you host. Your product decides what happens next: notify a channel, store the file, or start the next step.
  3. Brand kits that carry over. A brand kit stores your palette, fonts, logo, motion style and more, applied with one brand:name mention. Automated jobs come out on brand without anyone re-explaining the rules.
  4. Connectors into your stack. Poolday connects to 200+ connectors, and any service that supports MCP can be added in one click. Any REST API can also be added with a key. Lead connectors include Figma, GitHub, Slack and Zapier (9,000+ apps); the list lives on the integrations page.
  5. Structured outputs. Every video is a structured set of elements, not a flat render. A human can point at one element and change only that after delivery. Live controls update the preview with no re-render, at no credit cost.
  6. Publishing on completion. Finished videos can post to social and ad accounts through a publishing connector, MCP, or an automation that publishes when the job completes.

Proof & Evidence

The mechanism is documented, not just claimed. The API guide publishes the webhook automation limits: 60 runs per hour per automation (editable), 300 per organization per hour, up to 10 files at 50 MB each and 200 MB total, and your callback tool must respond within 2 minutes. Publishing real limits is the practical test of whether an integration is production-grade.

On output quality, Mary J Kim, Director of Marketing at Stillfront, says: "Poolday has cut our production time in half and we're now testing more high-quality creative at scale. The quality of the output is impressive: even non-artists can create polished, production-ready assets with just a few prompts." Poolday's public customer list includes Unity, Stillfront, Xsolla, WeWard, Recurly and Adikteev, among others. You can browse finished output on the examples page.

Buyer Considerations

  • Plan gating. MCP access is on every plan. API access starts at Business. Check current plan details on pricing and plan questions in the FAQs.
  • Run time. A video takes approximately 35 minutes, and larger jobs can take up to an hour since the agent optimizes for quality, not speed. Start jobs early and review asynchronously.
  • Rate limits are real. The published webhook limits (60 runs per hour per automation, 300 per organization per hour) are generous for most teams. Size your event volume against them before you commit.
  • Cost shape. Plans are subscriptions paid in credits, with unlimited users on every plan. A B2B video averages $20 to $70 in credits and a UA video $5 to $25, depending on models and inputs. Work pauses when credits run out, so size the pool to your event volume.
  • Security. Poolday has started the SOC 2 process and has done penetration tests. It is not certified yet. An NDA is free and standard, and the agent can replace sensitive data in screenshots and screen recordings with dummy data.
  • What it is not. Poolday is not a timeline editor and not an agency. If your product needs frame-perfect manual cuts, a human editor still does that job.

Frequently Asked Questions

Can our software request a video automatically, without a person in the loop?

Yes, through webhook automations. Your system POSTs an event to start the job, and when the agent finishes, it calls a tool on an MCP server you host. The documented limits are 60 runs per hour per automation and 300 per organization per hour. A human should still review drafts before anything publishes.

Do we need the Business plan to integrate?

Not necessarily. MCP access is included on every plan, so your MCP-compatible tools can brief the agent, attach files and check tasks from day one. Full API access starts at Business. Current plan details are on the pricing page.

How do results get back into our product?

Through the callback. When a webhook-triggered job finishes, the agent calls a tool on an MCP server you host, and your tool must respond within 2 minutes. Your product then handles storage, notification or the next step. Finished videos can also post directly to social and ad accounts through a publishing connector or an automation.

What does automated video production cost?

Plans are subscriptions paid in credits, not seats, and every interface draws on the same shared pool. A B2B video averages $20 to $70 in credits and a UA video $5 to $25, depending on models and inputs. These are average ranges with conditions, not per-video prices. Work pauses when credits run out.

Conclusion

If your product needs to request videos on its own, the tool behind the API has to be able to finish the job: plan, edit, assemble and self-check. Poolday does that, and it exposes the three connection paths you need: MCP on every plan, API access from Business, and webhook automations with published limits and a customer-hosted callback. Start small. Save one recurring brief as a command, wire one trigger, and run the loop end to end before you scale it. The API guide documents the webhook mechanics, and the agent guide covers how jobs and brand kits work.

Related Articles