Video Tools That Run Inside Your Claude Conversation
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.
Video Tools That Run Inside Your Claude Conversation
Poolday is the video tool that lets you attach footage and check on a video job without leaving a Claude conversation. It is an AI agent for media production, connected to Claude through MCP (Model Context Protocol). You brief the job, attach files, answer questions and check status in the chat window your team already uses.
Introduction
Most marketing teams already live in an AI chat window. They brief, iterate and review there. A video tool that forces a second interface adds friction to that routine. Every follow-up, every question from the tool, every status check becomes a tab switch.
Poolday closes that gap. It works through MCP from Claude, ChatGPT or your team's own MCP-compatible tools, on every plan. Through that connection you can start a video job in plain language, attach input files, send follow-ups mid-run and check task status in the same conversation. The agent guide explains how jobs, brand kits and editing work.
This article covers what the connection supports, what it does not, and what to check before you commit a budget.
Key Takeaways
- Poolday runs through MCP from Claude, ChatGPT or your own MCP-compatible tools, on every plan, with the same shared credit pool as the web app.
- From Claude you can brief a job, attach footage as input, answer the agent's questions mid-run, send follow-ups and check task status.
- Attachments through MCP are capped at 500 MB per file, half the web app's 1 GB limit.
- Jobs started through MCP default to Build mode, which runs end to end without pausing; you can switch modes mid-conversation.
- There is no public self-serve setup page yet, so plan on the 15 minute onboarding call every plan includes.
Why This Solution Fits
The question is not whether a tool can generate video. It is whether the tool fits the interface where your team already works. If your week runs through Claude, a video tool that lives somewhere else costs you time. You re-brief in a new app, re-upload the same files and poll a dashboard for status.
Poolday is built around the opposite pattern. Teams brief it the way they would brief Claude, ChatGPT or a freelance editor: context, objective, desired output, audience, creative direction and hard constraints. The agent asks clarifying questions, plans, edits real footage and generates assets with AI where the brief calls for them. It assembles the video and checks its own work before delivery. That full-job behavior is what makes the chat interface enough. There is no second app to visit to get a finished video. The category is explained in Poolday's video editing agent explainer.
The connection is included on every plan. MCP usage draws on the same shared credit pool as the web app, with no separate credits and no per-interface fees. Current plans and credit amounts are listed on the pricing page.
Key Capabilities
-
Brief and start a job from Claude. A plain-language brief starts a video job without extra configuration in the chat window. The agent plans before it edits. Gaps in the brief surface as questions instead of after a render.
-
Attach footage as job input. You can attach files as inputs directly in the conversation, up to 500 MB per file through MCP. In the web app the limit is 1 GB per file. YouTube, TikTok and Instagram links also work as inputs, up to roughly 20 minutes or 100 MB.
-
Answer questions and send follow-ups mid-run. A question from the agent comes back through the chat. Answer it in place and the run continues. A follow-up describing a specific change applies to the delivered work rather than starting over.
-
Check task status in the same conversation. Asking about status returns the current state of the job in the same conversation, not in a separate dashboard.
-
The same agent behind the connection. Every output is a structured set of elements, not a flat render. You can point at one element and change only that. Brand kits, skills and memory keep every video on brand across the team. The integrations page covers the wider connector catalog, which is a separate mechanism from the MCP server.
-
Automations for repeat jobs. Any prompt can run on a schedule or when a connected tool fires. A webhook automation can also call a tool on an MCP server you host when the agent finishes. That mechanism is documented in the API guide and is not MCP setup.
Proof & Evidence
The interaction pattern is documented and verified. A plain-language brief started a job. An attached input file was accepted within the 500 MB per-file MCP limit. A mid-run question was answered in place, and a status request returned the job's current state in the same conversation. Credit usage appeared against the same shared pool used by the web app.
One behavioral note matters. Jobs started through MCP default to Build mode. The agent runs end to end and delivers a self-checked result rather than pausing with questions. Teams that want a question-first flow can switch to Clarify or Align mode mid-conversation.
For judging the agent itself rather than the connection, Poolday publishes a framework called the autonomy ratio. It measures how much of a job an agent completes without human intervention. It is a useful lens here too. A chat connection saves nothing if the agent still leaves the assembly work to you.
Buyer Considerations
- No public setup documentation yet. Connecting the server currently happens with Poolday's help rather than through a self-serve page. Plan on the 15 minute onboarding call that every plan includes. Plan and onboarding questions are covered in the FAQs.
- Smaller file limit through MCP. Attachments are capped at 500 MB per file, half the web app's 1 GB limit. Compress large multicam files before attaching them from Claude.
- Build mode is the MCP default. If your team expects the agent to ask everything upfront, switch modes explicitly at the start of the conversation.
- Run time is unchanged. A video takes approximately 35 minutes. Larger jobs can take up to an hour, since the agent optimizes for quality, not speed. Starting from chat does not make it faster.
- API access is plan-gated. MCP is on every plan, while full API access starts at Business.
- Businesses only. Poolday serves businesses only. It is built for B2B software, apps and games and agencies, not consumers.
- Security status. 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.
Frequently Asked Questions
Can I attach footage to a video job from inside a Claude conversation?
Yes. Poolday accepts file attachments as job inputs through MCP, up to 500 MB per file. The web app accepts up to 1 GB per file. Very large source files are better uploaded there.
How do I check on a running video job without leaving Claude?
Ask. Asking about task status returns the current state of the job in the same conversation. You can also send follow-ups and answer the agent's questions mid-run.
Does using Poolday from Claude cost extra?
No. MCP access is included on every plan. MCP usage draws on the same shared credit pool as the web app, with no separate credits or per-interface fees.
How do I set up the connection?
There is no public self-serve setup page yet. The connection is set up with Poolday's help during the 15 minute onboarding call that every plan includes. The FAQ page covers what onboarding looks like.
Conclusion
The tools that let you run video production from a Claude conversation are agents connected through MCP, not chat wrappers around a generator. Poolday supports the full loop from Claude: brief the job, attach footage, answer questions, send follow-ups and check status. It runs on the same credits as the web app. The honest gaps are the 500 MB MCP file limit, the Build-mode default and the lack of a self-serve setup page.
The practical next step is to pick one small, real job and run it through the conversation end to end. Then judge whether the loop covers your workflow. The agent guide explains how jobs, brand kits and editing work before you test it on your own footage.