How YouTubers Plan and Organise a Content Pipeline With Claude or ChatGPT (2026): 8 Steps and 14 Prompts

Key Takeaways

• Storyflow is where a YouTube pipeline run with ChatGPT or Claude should live: through Storyflow's MCP server the assistant builds the idea bank, one wall per video with its script, shot cards and thumbnail references, and the edit brief, then reads back what your editor changed and who changed it.
• A chat thread is where you think, not where you keep a pipeline. Your editor cannot read your chat, and a new chat starts from nothing. The pipeline has to sit somewhere both people and the assistant can open.
• Run the pipeline as seven stages with a gate at each one. Most channels stall at two: an idea bank with no scoring, and packaging (title and thumbnail) done after the edit instead of before the script.
• Plan a month with your editor backwards from each publish date. In our worked example, 4 long videos and 8 Shorts need 2 batch shoot days, a footage handoff within 24 hours of each shoot, and a final cut 3 to 4 days before each upload.
• Use the assistant for three jobs: drafter (titles, scripts, briefs), clerk (moving the plan onto the board), and reviewer (the weekly "what changed and what is late" check).
• Notion, Airtable, ClickUp and Trello all have MCP servers that ChatGPT and Claude can use. They win when the pipeline is mainly a status list you sort, filter and date; Storyflow wins when each video needs its script, references and brief side by side on a canvas.
• Storyflow does not publish to YouTube, has no calendar view and cannot sort a table, so keep rows in date order and let YouTube Studio handle scheduling and analytics.
• The Storyflow MCP connection runs on your assistant's own model and has no action limit on Plus, Pro or Max; Storyflow is paid-only during early access, and an editor you invite joins free.

How Do YouTubers Plan and Organise Their Content Pipeline?

You have a list of video ideas in your notes app, half a script in a Google Doc, thumbnail references in a camera roll, and an editor asking on WhatsApp which video is next. That is not a pipeline. It is a pile. A pipeline is the same seven stages applied to every video, with a clear output at each stage and one place where you can see where every video sits.

The order matters more than the tool. Packaging comes second, not sixth: if you cannot write a title and a thumbnail concept that make you want to click, the script will not save the video. That is why the idea bank feeds packaging before it feeds the script, and why an idea that fails packaging goes back to the bank rather than forward to production.

The Status Layer and the Packet Layer

Every pipeline has two layers. The status layer answers "where is each video?" (Idea, Packaged, Scripted, Shot, In Edit, Scheduled, Live). The packet layer answers "what does the person working on this video need?" (the script, the shot list, the thumbnail references, the B-roll list, the music notes, the publish slot). Status tools such as Trello and ClickUp are built for the first layer. Most pipelines break on the second, because the packet ends up scattered across five apps and the editor asks the same questions every week.

What Changes When an Assistant Joins the Pipeline

An assistant does not add a stage. It changes who does the typing at each stage. In 2026 the useful split looks like this:

• Drafter. It writes 20 title variants, a first-draft script from your outline, a B-roll list from the script and an edit brief from your notes.
• Clerk. With a connector, it puts that work onto the board where the pipeline lives, so you are not copying from the chat into four apps.
• Reviewer. Once a week it reads the board and tells you what changed, what is late and what the editor is waiting on.

The second and third jobs only work if the assistant can reach the place the pipeline lives. That is what an MCP server does: it is the published connection that lets ChatGPT, Claude, Gemini or Grok read and write a tool on your behalf, after you sign in and pick what it may touch.

What ChatGPT and Claude Do Well in a Pipeline, and Where They Break

The familiar approach is to open a new chat for every video, paste in the topic, and work through titles and a script in the thread. That works for one video. It falls apart at the fourth, because the pipeline is spread across a dozen threads that neither your editor nor next week's chat can read.

What the Assistant Does Well

Assistants are fast at the stages that are mostly text. Ask for 20 title options in four angles (curiosity, outcome, contrarian, list) and you get them in under a minute. Ask for a script draft from a 10-bullet outline and you get a working draft you can rewrite in your voice. Ask for a B-roll list from a finished script and it pulls every visual the script implies, which is the list editors most often say they never receive.

Where the Plan Breaks

The chat cannot hold the pipeline over time. Three failures show up in the second week:

• The scroll. The idea bank you scored on Monday is 60 messages up by Thursday, and the assistant starts drifting from it.
• The fresh start. A new chat knows nothing about your channel, your last four videos or the video in edit, unless you paste it all again.
• The editor gap. Your editor works from what you send, not from your chat. If the edit brief stays in the thread, it does not exist for the person cutting the video.

The Split That Works: Draft in Chat, Hold on the Board, Measure in Studio

The working pattern is a three-way split. The assistant drafts. A board holds the pipeline, the packets and the dates. YouTube Studio publishes, tests and measures. Each tool does one job, and the assistant moves work between the first two. YouTube's own Test and compare feature runs up to three titles, thumbnails or title and thumbnail pairs on a long-form video and picks the winner on watch time, so the packaging options you draft on the board get a real test after upload.

How to Choose Where the Pipeline Lives: Our Criteria and Weighting

A pipeline tool for an assistant-run channel is judged on different things than a generic task manager. We picked six criteria from the jobs in the pipeline above, weighted them by how often each job comes up in a month, and scored five tools from 1 to 5 against each vendor's own documentation as of 9 October 2026.

Why the Packet Gets 30%

The pipeline breaks on the packet, not the status. A creator with one editor can track four videos in their head; what they cannot do from memory is hand over a full script, a B-roll list and three thumbnail references without something going missing. So the criterion that decides most of the score is whether the assistant can build the whole packet in the place the editor opens.

Why Change Awareness Is Scored at All

The weekly check is where an assistant earns its place. "What did the editor change in the last week?" is a question only the tool's change history can answer. If the connector cannot read that history, the assistant can only describe the board as it is now, and you lose the most useful weekly prompt.

What We Did Not Score

We did not score publishing, scheduling or analytics, because none of these five tools publishes to YouTube and YouTube Studio does that job for every channel. We did not score footage storage or frame-accurate review, because those belong to Frame.io, Google Drive or Dropbox. We did not score the assistant itself; that choice comes later in this guide.

Comparison Table: Where a YouTube Pipeline Run With ChatGPT or Claude Lives

Storyflow scores 2 on status and dates on purpose: it has no calendar view and cannot sort a table, and that is a real cost for a channel publishing daily. Trello scores 4 on change awareness because its MCP server can view action history and generate activity summaries, per Atlassian's documentation. We did not find a change-history tool in the Notion MCP pages we read, so Notion scores 2 there; if Notion adds one, its total rises to 4.0.

How to Run a YouTube Content Pipeline With Claude or ChatGPT in 8 Steps

These eight steps take a channel from scattered notes to a pipeline the assistant can work in. Steps 1 and 2 are one-time setup (about 40 minutes). Steps 3 to 8 repeat every month.

Step 1: Connect the Assistant and Pick What It May Touch

Connect Storyflow to Claude or ChatGPT with the address https://mcp.storyflow.so/mcp (exact clicks are in the connection section below). At sign-in, pick only the project that holds your channel, and allow changes, because a read-only connection cannot build the board. If you use a freelance editor who also works with an assistant, they connect their own account with their own permissions; the assistant never shares your sign-in.

Then make one empty project called by your channel name, with one board called Pipeline. Everything in this guide lands there. If you would rather start from a ready layout, the YouTube content pipeline board shows the stages laid out on a canvas.

Step 2: Write the Channel Brief Once, on the Board

The assistant plans better from a short brief than from a long chat history. Write five lines and ask the assistant to put them on the board as the first wall:

• Who the channel is for, in one sentence.
• The 3 or 4 formats you make (for example: 20-minute explainer, 8-minute review, Shorts cut from the long video).
• Your publish rhythm (for example: long video every Tuesday, Shorts Thursday and Saturday).
• The two numbers you care about this quarter (for example: average view duration above 45%, click-through rate above 6%).
• What you will not make.

Every later prompt says "read the Channel Brief wall first". That one habit removes most of the generic output people complain about.

Step 3: Build the Idea Bank With a Scoring Gate

An idea bank without a gate becomes a graveyard. Ask the assistant to build an Ideas table on the board with five columns: idea, demand (1 to 5), packaging strength (1 to 5), cost to make (5 is lowest cost), fit with the brief (1 to 5), and a total out of 20. Paste in your raw ideas and ask it to score them, then rescore anything you disagree with yourself. Ideas at 14 or more move to packaging; the rest stay in the bank. The gate is not the assistant's judgment. It is your judgment, written down so the assistant can apply it consistently.

For demand, check what already works in your niche with YouTube search and your own analytics. The assistant can suggest angles, but it cannot see your channel's search traffic unless you paste the numbers in.

Step 4: Package Before You Script

For every idea that passed the gate, ask for 10 title options in four angles, then pick 3. Ask for 3 thumbnail concepts written as a picture description, a two-to-four word text overlay and the one feeling the image should create. Put the three titles and the concepts on the video's wall, with any reference thumbnails beside them. If you have a reference image, add it with an upload link: a picture attached in the chat does not reach Storyflow, so the assistant gives you a link to upload it to the board.

If none of the three titles makes you want to click, the idea goes back to the bank, before anyone films. For more on concepts that earn the click, see the AI thumbnail concept generator.

Step 5: Script and Shot Plan per Video

With packaging fixed, ask the assistant to write the script into a document on that video's wall: a hook that pays off the title within 30 seconds, chapters, and a retention reset (a new question, a reveal or a change of scene) every 60 to 90 seconds. Rewrite it in your voice; the assistant's draft is a skeleton, not the final words. Then ask for shot cards from the script (one card per setup) and a B-roll list as a checklist. Storyflow's MCP server can build shot cards directly, which is the fastest way to turn a script into a shoot plan.

Lock the script 7 days before the batch shoot. A script that changes on the shoot day is the most common reason an editor receives footage that does not match the brief.

Step 6: Plan the Month With Your Editor, Backwards From Each Publish Date

This is where "how do content creators plan a month of videos with their editor?" gets answered. Start from the publish dates and work back:

• Publish date. Fixed by your rhythm.
• Final cut. 3 to 4 days before publish, so you have time for the thumbnail, the description and one fix.
• Notes on the first cut. One round, returned within 24 hours.
• First cut. 2 to 5 working days after the editor receives footage, depending on length and what else is in the edit.
• Footage handoff. Within 24 hours of the shoot.
• Batch shoot. Two videos per shoot day for most talking-head and explainer formats.

Hold a 60 to 90 minute planning call with your editor once a month. Share the board, agree the four publish dates, and let the assistant write the backwards schedule into a Schedule table on the board, rows kept in date order. The editor sees their dates beside the packets they will cut, not in a separate message.

Step 7: Hand Off One Packet per Video

The edit brief is the single most useful document an assistant writes in this pipeline. Ask it to write one per video, into a document on that video's wall, with these parts: the goal of the video in one sentence, the hook and where it pays off, the chapter list with timestamps from the script, the B-roll list, the tone and pacing reference (link one video the editor should match), music notes, the title and thumbnail it has to deliver on, delivery specs and the deadline. Then invite the editor to the board. They join free, and see the brief beside the script, the references and the shot cards. The video editing brief page shows the same structure as a standalone board, and Storyflow for video editors covers the editor's side.

Footage still goes through Google Drive, Dropbox or Frame.io. Put the folder link on the video's wall so the editor finds it in the same place every time.

Step 8: Run the Weekly Check and the Month-End Review

Once a week, ask the assistant: "Read my Pipeline board. Tell me what changed in the last week and who changed it, which video is behind its schedule row, and what the editor is waiting on from me." Storyflow's MCP server can read what changed on a board and who changed it, so the answer reflects the editor's work, not only yours.

At month end, paste the numbers from YouTube Studio for each video (views at 7 days, average view duration, click-through rate, the best and worst comment theme) onto the board, and ask the assistant which packaging and which format moved your two numbers. Change next month's slate by one or two videos, not all four.

The 14 Prompts to Copy

Copy these as they are, then replace the parts in capitals. Each assumes Storyflow is connected and your channel brief is on the board.

Prompts for the Weekly Check

Prompts for the Month-End Review

A Worked Example: November for a Tech-Explainer Channel and One Editor

This example is illustrative: a solo creator running a tech-explainer channel with one freelance editor, planning November 2026 in mid-October. The numbers are the ones we would plan with; your formats will shift them.

The Brief and the Numbers

The channel brief on the board says: explainers for people who buy their own laptops and phones; long video every Tuesday, Shorts every Thursday and Saturday; two numbers for the quarter, average view duration above 45% and click-through rate above 6%; no unboxings. The goal for November is 4 long videos and 8 Shorts cut from them.

The creator pastes 40 raw ideas. The assistant scores them; the creator rescores 9. Eleven ideas reach 14 or more out of 20. Packaging kills 4 of those, because none of the three titles worked. Seven survive, and the top four become November's slate. The other three stay packaged and ready for December.

The Slate

The Production Calendar, Built Backwards

Two shoot days cover four videos. The editor always has one video in first cut and one waiting, never three at once, and every final cut lands on a Friday with a weekend of slack before the Tuesday upload.

What the Assistant Built on the Board

Over the month, the assistant built and maintained six walls: Channel Brief, Ideas (a 40-row scored table), one wall per video (three titles, three thumbnail concepts, the script document, shot cards, a B-roll checklist, the edit brief and the footage folder link), and a Schedule table. It added 8 Shorts tickets, two per video, each with a start line, an end line and an overlay.

Counting every read and every change, that came to about 120 MCP actions in the month: roughly 45 to build the walls, 35 for scripts and briefs, 24 across four weekly checks and the rest for fixes. On Plus, Pro or Max there is no MCP limit, so the count does not matter. If the editor connects their own assistant from a free invited account, they get 25 actions a week, which is enough for reading their briefs and the weekly change summary, though not for building the pipeline itself.

What the Month Took

About 32 hours of the creator's time for 4 long videos and 8 Shorts, of which roughly 5 hours are planning and admin that used to happen in messages. The time the assistant saves is not in the script, which still needs your rewrite. It is in the briefs, the schedule and the weekly status questions that no longer need a call.

How to Connect Claude or ChatGPT to Storyflow

Connecting takes about two minutes and needs no code. Every assistant uses the same address: https://mcp.storyflow.so/mcp. The help article on connecting AI assistants has the steps for every client, and the Storyflow MCP page lists what the assistant can read and build.

Connect Claude

1. In Claude, open Customize, then Connectors.
2. Click Add, then Add custom connector.
3. Name it Storyflow, paste the address and click Continue.
4. Choose Sign in now, click Add, and sign in to Storyflow when asked.

Anthropic's custom connector guide says custom connectors work on Free, Pro, Max, Team and Enterprise, and that Free users are limited to one custom connector. On Claude Team and Enterprise, an owner adds the connector for the organisation first. Claude Code users run: claude mcp add --transport http storyflow https://mcp.storyflow.so/mcp, then type /mcp to sign in.

Connect ChatGPT

1. In ChatGPT on the web, open Settings, then Security and login, and turn on Developer mode. On Business and Enterprise, an admin turns it on.
2. Go to chatgpt.com/plugins and click the plus button.
3. Name it Storyflow, choose Public endpoint and paste the address.
4. Create it and sign in to Storyflow when asked.

Custom connectors in ChatGPT need ChatGPT Plus or above. OpenAI's developer mode article covers the mode itself; expect ChatGPT to ask you to confirm before the assistant first writes to your board.

The Permissions to Pick at Sign-In

Storyflow asks three things when you sign in. Pick the channel's project rather than all projects, so the assistant cannot reach client work or other channels. Allow changes, or it can only read. Decide whether it may use documents outside projects; this starts off when you pick projects. Unknown apps show an "I trust this app" warning first. You can disconnect in Settings, then Connected apps, and access ends within a minute. Connections expire after 30 days, and you sign in again.

No Outside Assistant? Use Storyflow's Own AI

You do not need ChatGPT or Claude for this. Storyflow's built-in AI has a YouTube video plan workflow that plans one video for a channel that keeps publishing, across eight walls: theme and overview, research and facts, chronology, the formats that work on your channel, production, assets, distribution with the publish slot and channel metrics, and budget and risk. A video script workflow writes a three-act script with the production around it, and a video research workflow builds a research board for one subject. The built-in AI reads the current board plus up to five attached documents and three other boards. It uses your Storyflow AI allowance; the MCP connection runs on your assistant's model and does not spend Storyflow AI credits.

Claude or ChatGPT: Which Assistant for a YouTube Pipeline?

Both work with the same Storyflow board, so the choice is about how you like to write, not about the pipeline. Pick one as the main assistant for a month so your prompts and habits stay consistent.

Pick Claude If

You write long scripts and want a draft that keeps your structure, you want to edit a long document in one pass, or you want a connector on the free plan (Claude Free allows one custom connector). Claude's pricing page lists Pro at $20 a month, or $17 a month billed annually, and marks connectors as available on Free.

Pick ChatGPT If

You already pay for ChatGPT Plus or above, you want image ideas for thumbnail concepts in the same chat, or your editor already uses ChatGPT. Developer mode is needed to add Storyflow as a custom connector, and on Business or Enterprise an admin enables it.

Use Both Without Losing the Pipeline

You can run Claude for scripts and ChatGPT for packaging, because the board is the shared memory. Both read the same Channel Brief wall and the same video walls. The board, not either chat, is the source of truth, so the rule is simple: anything you want to keep goes on the board before you close the chat.

Notion, ClickUp, Trello and Airtable With an Assistant: What Each Pipeline Looks Like

Every tool in the comparison table can be driven by ChatGPT or Claude in October 2026. What differs is the shape of the pipeline the assistant builds, and what your editor sees when they open it.

1. Storyflow

Verdict: the pipeline as a canvas, one wall per video with its packet beside its status. Best for: creators with an editor who need the script, references, shot cards and brief in one place. Pricing: Plus $7.99 a month billed annually ($9.99 monthly), Pro $14 ($19 monthly) adds AI image generation, Max $39 ($49 monthly) adds a team workspace with roles; paid-only during early access, with a Free plan launching before the end of 2026, and invited collaborators join free.

Through the MCP server the assistant builds notes, tables, documents, shot cards, mind maps, charts and slide decks, applies board templates, edits and moves elements, and reads what changed on a board and who changed it. What it adds appears live on an open board.

• Strengths: the packet layer, packaging references beside titles, change summaries for the weekly check, no MCP limit on paid plans.
• Limits: no calendar view, no table sorting, no YouTube publishing or analytics, browser only.

2. Notion

Verdict: the pipeline as a database, one page per video. Best for: teams already in Notion who want filters and a calendar of publish dates. Pricing: Notion's pricing page lists Free, Plus at $10 per member a month billed annually ($12 monthly) and Business at $20 billed annually ($24 monthly); full Notion AI is on Business.

Notion MCP is a hosted server that lets an assistant search, read, create and update pages and databases. Notion's help article notes that tools act with your full Notion permissions and that it is designed for individual workflows first.

• Strengths: views, filters, relations, a mature template ecosystem.
• Limits: the packet becomes a long page, references sit below the fold, and we found no change-history tool in the MCP pages we read.

3. Trello

Verdict: the simplest status board, with a surprisingly good weekly check. Best for: a creator and one editor who need a card per video and little else. Pricing: Trello's pricing lists Free for up to 10 collaborators per workspace, Standard at $5 per user a month billed annually ($6 monthly) and Premium at $10 ($12.50 monthly).

The Trello MCP server lists ChatGPT, Claude, Cursor, VS Code and Gemini CLI. It can create and move cards, manage checklists and comments, and view action history and generate activity summaries. It archives rather than deletes, and cannot upload local files.

• Strengths: free for two people, fast status moves, activity summaries.
• Limits: a script or an edit brief lives in a card description, and thumbnail references are attachments you click into.

4. ClickUp

Verdict: the pipeline as a task system with owners and due dates. Best for: channels run like a small studio, with several people and many dates. Pricing: ClickUp's pricing page lists a free plan with unlimited free members; on 9 October 2026 it showed two versions of its paid prices, so check the current one.

ClickUp's MCP setup guide uses https://mcp.clickup.com/mcp and covers Claude, Claude Code, ChatGPT and Cursor; in ChatGPT you connect the ClickUp app from Apps.

• Strengths: dates, dependencies, workload views, docs inside tasks.
• Limits: more setup than a two-person channel needs, and the packet still splits between a task and a doc.

5. Airtable

Verdict: the pipeline as structured data. Best for: channels that track many videos across several shows and want reporting. Pricing: Airtable's pricing lists Free, Team at $20 per seat a month billed annually and Business at $45.

Airtable's MCP server is available on all plans, works with Claude, ChatGPT and Gemini, and can read and write records, record comments and schema, within your base permissions.

• Strengths: fields, formulas and interfaces for a channel with volume.
• Limits: the most expensive seat for an editor, and creative references sit in attachment fields rather than beside each other.

What a YouTube Pipeline With AI Costs: Creator Plus One Editor

Prices as listed on each vendor's page on 9 October 2026, monthly, for a creator and one editor. The assistant is a separate cost and does not change between rows.

Storyflow's case on cost is not that it is the lowest; Trello Free is lower. Current Storyflow prices are on the pricing page. It is that one $7.99 plan covers the packet, the board and an editor who joins free, with no MCP limit. Frame.io's pricing is listed for channels that want frame-accurate review comments on cuts; its Free tier covers 2 members and 2GB.

Which Setup for Which YouTuber

Decision Rules

• If your editor keeps asking for the script, the B-roll list or the references, the problem is the packet: use Storyflow.
• If you publish more than 10 pieces a week and need to sort and filter, use Notion, ClickUp or Airtable for status, and keep packets for long videos on a Storyflow board if you want them.
• If you pay nothing today and have one editor, start with Trello Free and move when the packets get lost.
• If you want the assistant to tell you every week what the editor changed, use Storyflow or Trello, whose MCP servers can read change history.
• If YouTube Studio already tells you what you need, do not build a dashboard in your pipeline tool; paste the 7-day numbers in once a month.

Common Mistakes When ChatGPT or Claude Runs Your Pipeline

These are the failures that show up most often once an assistant is in the loop. Each has a one-line fix.

Mistake 1: Keeping the Pipeline in the Chat

The idea bank, the scores and the scripts sit in a thread the editor cannot read and the next chat cannot see. Fix: anything you want next week goes on the board before you close the chat.

Mistake 2: Letting the Assistant Score Ideas Alone

An assistant scores demand from what sounds popular, not from your channel's numbers. Fix: write your scoring rules on the board, paste your own analytics, and rescore anything that feels wrong.

Mistake 3: Packaging After the Edit

A title written after the edit describes the video you made, not the click you need. Fix: three titles and three thumbnail concepts before the script, and the idea goes back to the bank if none of them works.

Mistake 4: Publishing the Assistant's Script Unedited

Viewers hear a draft voice in the first 30 seconds. Fix: use the draft for structure and the retention resets, then rewrite every line you will say on camera.

Mistake 5: Sending Footage Without a Brief

The editor guesses the pacing, the B-roll and the hook. Fix: one edit brief per video, written by the assistant from your notes and placed on the video's wall before the footage arrives.

Mistake 6: Giving the Connector Every Project

An assistant with access to every project can write into the wrong board. Fix: at sign-in, pick only the channel's project, and reconnect if you need another.

Mistake 7: Skipping the Month-End Review

Without it, next month's slate repeats this month's guesses. Fix: paste the 7-day numbers for each video onto its wall and ask which title angle and format moved your two numbers.

When Storyflow Is Not the Right Choice

Storyflow fits a pipeline where each video needs a packet the editor can read. It is the wrong tool in these cases:

• You need a calendar view or sorted lists. Storyflow has no calendar view and cannot sort a table. A channel planning 40 Shorts a month by date is better served by Notion, ClickUp or Airtable.
• You want the pipeline tool to publish or schedule. Storyflow does not upload to YouTube or schedule posts. YouTube Studio does that, and nothing in this guide changes it.
• You need frame-accurate review with an approve button. Storyflow has no versioned proofing; use Frame.io for cut review.
• You work offline or on a phone first. Storyflow runs in the browser, with no native mobile or desktop app and no offline mode.
• You want a free tool today. Storyflow is paid-only during early access; its Free plan launches before the end of 2026. Trello Free covers a creator and an editor at no cost.
• Your team has standardised on another tool. If your studio runs everything in ClickUp or Notion, the switching cost may outweigh a better packet; connect the assistant to the tool you have.

A Pipeline Board Template You Can Copy

Build this once, by hand or by asking the assistant to build it from this section. Every month after that, you add four video walls and a new schedule.

The Six Walls

1. Channel Brief. Audience, formats, rhythm, two numbers, what you will not make.
2. Ideas. The scored table, highest totals first.
3. This Month. The slate: four video titles with their publish dates.
4. Video walls, one per video. Titles, thumbnail concepts and references, script document, shot cards, B-roll checklist, edit brief, footage link, 7-day numbers after publish.
5. Schedule. The backwards production table, rows kept in date order.
6. Review. One note per month: what moved the numbers and what changes next month.

The Video Wall Checklist

The Edit Brief Outline

Goal in one sentence. Hook and where it pays off. Chapters with timestamps. B-roll list. Pacing reference (one linked video). Music notes. Title and thumbnail the video must deliver on. Delivery specs (resolution, captions file, Shorts crops). Deadline for first cut and final cut.

The Monthly Checklist

• Score new ideas and package the top 7.
• Pick 4 and hold the planning call with the editor.
• Build the schedule backwards from the publish dates.
• Lock scripts 7 days before each batch shoot.
• Write every edit brief before the footage arrives.
• Run the weekly check every Monday.
• Paste the 7-day numbers and run the month-end review.

The Bottom Line

A YouTube pipeline is seven stages and a packet per video, and the assistant's job is to draft that work and move it to where your editor can see it. Run ChatGPT or Claude for the drafting and keep the pipeline on a Storyflow board through its MCP server: the scored ideas, one wall per video with its script, references and edit brief, a schedule built backwards from publish dates, and a weekly check that reads what your editor changed. If your channel is a high-volume list you sort by date, Notion, ClickUp or Airtable will serve you better. For everyone else, connect your assistant on the Storyflow MCP page, build the six walls above, and plan next month in one call with your editor.