Showing 31 of 31 playbooks · editorial examples based on linked product documentation
Editorial playbook·Analyst·20–40 min·Intermediate
Create a recurring weekly report in a ChatGPT Project
Repeatable reporting using the documented Projects capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for create a recurring weekly report in a chatgpt project.
- 2Use the documented Projects capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·20–45 min·Intermediate
Review a document set and extract action items with ChatGPT
Turn files into actions using the documented Files capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for review a document set and extract action items with chatgpt.
- 2Use the documented Files capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·10–25 min·Intermediate
Answer a current question with ChatGPT Search and citations
Find current evidence using the documented Search capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for answer a current question with chatgpt search and citations.
- 2Use the documented Search capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·10–25 min·Intermediate
Inspect a chart or screenshot and turn it into next steps
Understand visual information using the documented Vision capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for inspect a chart or screenshot and turn it into next steps.
- 2Use the documented Vision capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–60 min·Intermediate
Create a focused custom GPT from instructions and knowledge files
Build a reusable assistant using the documented Custom GPTs capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for create a focused custom gpt from instructions and knowledge files.
- 2Use the documented Custom GPTs capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·20–45 min·Intermediate
Draft and revise a document in ChatGPT Canvas
Improve a working draft using the documented Canvas capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for draft and revise a document in chatgpt canvas.
- 2Use the documented Canvas capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–60 min·Intermediate
Build a support playbook from approved documents in Claude
Standardize support answers using the documented Projects capability in Claude
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for build a support playbook from approved documents in claude.
- 2Use the documented Projects capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–60 min·Intermediate
Turn product requirements into a project plan with Claude
Plan product work using the documented Projects capability in Claude
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for turn product requirements into a project plan with claude.
- 2Use the documented Projects capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–90 min·Intermediate
Run multi-search research and produce a cited brief with Claude
Research a decision using the documented Research capability in Claude
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for run multi-search research and produce a cited brief with claude.
- 2Use the documented Research capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–60 min·Intermediate
Use Claude Project knowledge for a large document set
Query a large knowledge base using the documented Projects capability in Claude
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for use claude project knowledge for a large document set.
- 2Use the documented Projects capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·45–90 min·Intermediate
Compare market sources and record open questions with Claude Research
Scan a market using the documented Research capability in Claude
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for compare market sources and record open questions with claude research.
- 2Use the documented Research capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–60 min·Intermediate
Share a reviewed project workspace with a team
Collaborate on shared context using the documented Projects capability in Claude
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for share a reviewed project workspace with a team.
- 2Use the documented Projects capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Official documentation·Analyst·15–30 min·Intermediate
Return structured JSON from a DeepSeek review task
Create machine-readable output using the documented JSON output capability in DeepSeek
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for return structured json from a deepseek review task.
- 2Use the documented JSON output capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Official documentation·Analyst·20–45 min·Intermediate
Stream a DeepSeek response into an application interface
Build responsive AI UX using the documented Streaming capability in DeepSeek
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for stream a deepseek response into an application interface.
- 2Use the documented Streaming capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Official documentation·Analyst·30–60 min·Intermediate
Use DeepSeek function calling for an approved application action
Connect AI to tools using the documented Function tools capability in DeepSeek
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for use deepseek function calling for an approved application action.
- 2Use the documented Function tools capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Official documentation·Analyst·20–45 min·Intermediate
Manage conversation state explicitly with the DeepSeek Responses API
Build multi-turn API flows using the documented Responses API capability in DeepSeek
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for manage conversation state explicitly with the deepseek responses api.
- 2Use the documented Responses API capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Official documentation·Analyst·15–30 min·Intermediate
Tune DeepSeek reasoning effort for speed versus depth
Balance quality and latency using the documented Thinking mode capability in DeepSeek
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for tune deepseek reasoning effort for speed versus depth.
- 2Use the documented Thinking mode capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Official documentation·Analyst·30–60 min·Intermediate
Validate DeepSeek tool arguments before execution
Add safe automation controls using the documented Tool calling capability in DeepSeek
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for validate deepseek tool arguments before execution.
- 2Use the documented Tool calling capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–90 min·Intermediate
Turn scattered source material into a finished deliverable with ChatGPT Work
Finish a multi-step task using the documented Work capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for turn scattered source material into a finished deliverable with chatgpt work.
- 2Use the documented Work capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Editorial playbook·Analyst·15–30 min·Intermediate
Set up a recurring ChatGPT task for a reminder or check
Keep a recurring task on track using the documented Scheduled tasks capability in ChatGPT
Goal: Apply AI to a real task
- 1Define the task, audience, inputs, expected output, and review criteria for set up a recurring chatgpt task for a reminder or check.
- 2Use the documented Scheduled tasks capability and provide only the files, sources, permissions, or tools needed for the task.
- 3Review the output, assumptions, citations or tool results, then correct errors before sharing or automating anything.
- + 1 more steps
View workflow →Official documentation·Analyst·30–90 min·Intermediate
Run a cited research report with ChatGPT Deep Research
A structured report that synthesizes selected sources with citations
Goal: Research with citations
- 1Describe the research outcome, audience, scope, date range, and required report structure. Add relevant files or connected sources when available.
- 2Review and refine ChatGPT's proposed research plan, including the websites and source context it should use.
- 3Let Deep Research investigate and synthesize the sources, then review the cited findings, disagreements, assumptions, and evidence quality.
- + 1 more steps
View workflow →Official documentation·Analyst·20–60 min·Intermediate
Analyze a spreadsheet and create a reviewed data story with ChatGPT
A cleaned analysis with tables, charts, assumptions, and next actions
Goal: Get insights faster
- 1Upload a structured CSV, spreadsheet, PDF, or supported data file with clear column names and one record per row.
- 2State the questions, calculations, groupings, date range, and output format you need, then ask ChatGPT to inspect data quality first.
- 3Review the generated tables, charts, Python calculations, assumptions, missing values, and outliers before relying on the findings.
- + 1 more steps
View workflow →Official documentation·Analyst·30–60 min·Intermediate
Build a reusable research workspace with Claude Projects
A focused project workspace that turns source documents into repeatable research outputs
Goal: Organize research
- 1Create a focused Claude Project for one research topic and add a short description of the intended outcome.
- 2Upload authorized documents to the project knowledge base and add project instructions for source priority, audience, tone, and uncertainty handling.
- 3Ask Claude to map the sources and summarize the strongest evidence before asking it to form conclusions.
- + 1 more steps
View workflow →Official documentation·Developer·20–45 min·Advanced
Use DeepSeek thinking mode with tool calls for a reviewed task
A structured answer that can use approved functions before producing a final result
Goal: Build reliable AI tasks
- 1Define the task, expected output, approved tools, and safety limits. Remove secrets and private data before sending inputs.
- 2Enable DeepSeek thinking mode and provide the available function tools with clear names, descriptions, and parameter schemas.
- 3When DeepSeek requests a tool call, validate the arguments, run only the approved function, and pass the tool result back with the required reasoning content.
- + 1 more steps
View workflow →Editorial playbook·Founder / CEO·30–45 min·Intermediate
Analyze SaaS signups and subscription plans without SQL
A reusable workflow for weekly signup analysis, plan comparison, dashboards, and recurring reporting
Goal: Understand SaaS growth
- 1Connect a safe sample or authorized database containing user registration and subscription data. Illustrative logical tables are users (id, created_at, email) and subscriptions (id, user_id, plan_name, status, created_at); these fields are examples, not verified product requirements. Use read-only credentials where possible.
- 2Ask: "How many users signed up each week for the last 8 weeks?" Review the generated query for the selected table or collection, signup timestamp, time window, weekly grouping, timezone assumptions, and counts against a trusted sample or manual check.
- 3Ask: "Which plan do new users choose most?" Review the relationship between users and subscriptions, the signup period, free/trial/paid interpretation, multiple subscriptions per user, and cancelled subscriptions.
- + 2 more steps
View workflow →Official documentation·Analyst·30–60 min·Intermediate
Turn a project knowledge base into a reviewed decision brief with Claude
A reusable project workspace and decision brief grounded in the documents you provide
Goal: Make better decisions
- 1Create a focused Claude Project for one topic, such as a product launch, research question, or customer-support playbook. Projects and project knowledge are available according to the current Claude plan and workspace settings.
- 2Upload only documents you are authorized to use, then add project instructions describing the audience, source priority, tone, output format, and how Claude should flag missing or conflicting evidence.
- 3Ask Claude to build a source map that lists each document's purpose, date, owner, and key claims. Check the map against the uploaded files before asking for conclusions.
- + 2 more steps
View workflow →Official documentation·Developer·20–40 min·Intermediate
Use DeepSeek reasoning to review a technical proposal
A structured technical review with assumptions, risks, and follow-up actions
Goal: Review technical proposals
- 1Prepare a non-sensitive technical proposal, requirements document, or code excerpt and define the decision you need to make. Remove secrets, private customer data, and credentials before sharing it.
- 2Ask DeepSeek to identify the proposal's assumptions, missing requirements, failure modes, and trade-offs. Request a structured response with evidence from the supplied text rather than unsupported certainty.
- 3Ask for two or three viable alternatives and a comparison table covering complexity, cost, reliability, security, and migration effort. Mark unknowns that require human or external verification.
- + 2 more steps
View workflow →Editorial playbook·Creator·25–40 min·Beginner
Turn one article into a social carousel
A short, editable carousel based on your own article
Goal: Create content faster
- 1Choose one article you own and identify one useful takeaway for a specific audience.
- 2Ask ChatGPT for a 6-slide outline: hook, four practical points, and a final call to action. Check every claim against the original article.
- 3Use Canva Magic Design to draft a carousel layout, then replace its text with your reviewed outline and apply your brand assets.
- + 1 more steps
View workflow →Editorial playbook·Creator·20–35 min·Beginner
Make a captioned short from a recorded video
One short video with manually checked captions
Goal: Repurpose video
- 1Pick a recording you have permission to edit and find one self-contained 30–60 second idea.
- 2Trim the recording in CapCut and place the strongest sentence near the start.
- 3Use Auto Captions, select the spoken language, and generate subtitles.
- + 1 more steps
View workflow →Editorial playbook·Designer·20–30 min·Beginner
Create a branded social visual from a brief
A reviewed image post with a clear message
Goal: Create branded visuals
- 1Write a brief with audience, message, visual style, format, and any words that must appear exactly.
- 2Generate or edit an image in ChatGPT Images; inspect details and save the chosen version.
- 3Bring the image into Canva Magic Design, choose a suitable layout, then add your real logo and reviewed copy.
- + 1 more steps
View workflow →Editorial playbook·Analyst·30–45 minutes·Intermediate
Analyze SaaS Signups and Subscription Plans Without SQL
A reusable, reviewable workflow for understanding weekly SaaS signups, popular subscription plans, dashboards, and recurring reporting without writing SQL. Do not fabricate counts or growth percentages; actual results depend on the connected dataset and current product capabilities.
Goal: Apply AI to a real task
- 1Prepare a safe sample or authorized database containing user registration and subscription data. Illustrative logical tables are users (id, created_at, email) and subscriptions (id, user_id, plan_name, status, created_at); these fields are examples, not verified product requirements. Connect the database with read-only credentials where possible.
- 2Ask: "How many users signed up each week for the last 8 weeks?" Review the generated query and check the selected users table or collection, signup timestamp, eight-week window, weekly grouping, timezone assumptions, and counts against a trusted sample or manual check.
- 3Ask: "Which plan do new users choose most?" Review the generated query for the users-to-subscriptions relationship, signup period, treatment of free, trial, and paid plans, multiple subscriptions per user, and cancelled subscriptions.
- + 2 more steps
View workflow →Editorial playbook · 25–40 min
Turn one article into a social carousel
- Choose one article you own and identify one useful takeaway for a specific audience.
- Ask ChatGPT for a 6-slide outline: hook, four practical points, and a final call to action. Check every claim against the original article.
- Use Canva Magic Design to draft a carousel layout, then replace its text with your reviewed outline and apply your brand assets.
- Read every slide on a phone, fix cramped text, and export the final design.