ChatGPT Work Practical Guide: 6-Role Workflow Prompt Templates + Scheduled Tasks Recipes (2026)

Intro: After OpenAI launched ChatGPT Work on July 9, 2026, the real question is: what can you actually use it for tomorrow morning? OpenAI's advice — hand it a task you already know well. This guide follows that path, breaking down real workflows across Sales, Marketing, Finance, Operations, Product, and Engineering with copy-paste prompt templates, Plan Mode review checklists, Scheduled Tasks automation recipes, and usage optimization tips. For launch context and Cowork comparison, read the companion post: ChatGPT Work Launch Deep Dive.

1. Pain Points: Why You Know About Work But Can't Use It

Wrong mode wastes usage — simple Q&A in Work, complex cross-app tasks in Chat: costs can differ by 5×.
Prompts describe steps, not outcomes — Work plans its own path; you only need to specify the deliverable.
Scheduled Tasks without safety checks — running unattended without limiting plugin scope or disabling auto-send is high risk.

2. Before You Start: 3 Principles That Decide Success

Principle	What It Means	Practical Tip
Describe outcomes, not steps	Work plans its own path	❌ "Open Salesforce and export…" → ✅ "From @Salesforce last-30-day opps, generate a weekly PPT with risk flags"
Connect tools first, then assign tasks	Plugin directory is your data source	Confirm Gmail, Slack, Drive are authorized; use `@AppName` explicitly
Plan Mode is your brake	Complex tasks: plan first, execute second	Review line-by-line before external emails, financial reports, or client deliverables ship

2.1 Pick the Right Mode: Chat / Work / Codex Quick Routing

Your Need	Recommended Mode	Why
Quick Q&A, brainstorming, single-turn copy	Chat	Lightweight, fast response
Cross-app multi-step tasks, finished deliverables, multi-hour runs	Work	Plugins + Plan Mode + Computer Use
Code review, PR management, multi-repo development	Codex	Developer-specific workflows
Weekly recurring, unattended background tasks	Work + Scheduled Tasks	Scheduled/triggered autonomous execution

2.2 Desktop vs Web: Which Environment to Use

Scenario	Recommended Environment
Local file read/write, Computer Use, free-tier trial	Desktop (Mac / Windows)
Team collaboration, checking task progress on the go	Web / Mobile (Plus and above)
Sales meeting brief auto-generation + email notification	Web Workspace Agent + scheduling
Local Excel reconciliation, folder batch processing	Desktop Work mode

3. Universal Workflow: 5 Steps to Run Your First Task

1. Connect plugins → 2. Define goal and output format → 3. Review Plan Mode → 4. Intervene mid-run to correct → 5. Accept deliverable and iterate

3.1 Work Mode Prompt Formula

[Role] + [Data source @plugin] + [Specific task] + [Output format] + [Constraints] + [Acceptance criteria]

Template skeleton:
You are a [role]. Pull [data type] for [time range] from @Salesforce and @Gmail.
Complete [specific action], output as [Google Docs / Excel / PPT / Sites].
Constraints: [do not modify source data / amounts to 2 decimal places / do not send external emails].
When done, [Slack notify me / save to specified folder].

3.2 Plan Mode Review Checklist

Are data sources correct (right customer / month)?
Any high-risk actions — external send, delete, file overwrite?
Does output format match team templates?
Can intermediate steps be trimmed to save usage?
Do you need a human confirmation checkpoint?

4. Six-Role Practical Workflows (With Prompt Templates)

Templates below are adapted from OpenAI official examples, early feedback from Zapier / Nvidia / Virgin Atlantic, and the Workspace Agent Cookbook. Replace @PluginName with your actual tool stack.

4.1 Sales

Scenario A: Auto Customer Meeting Brief (Daily Scheduled) — Pain point: reps spend 1–2 hours daily prepping client context. Work solution: scan calendar → pull CRM → search news → generate brief.

Create a scheduled task: run every weekday at 4:00 PM.

1. Check my @Google Calendar for tomorrow's client meetings (exclude internal meetings)
2. For each client meeting:
   - Pull last-30-day account notes and interactions from @SharePoint / @Salesforce
   - Search public news and executive activity for that company in the last 30 days
   - Write a 2–3 sentence background summary for each external attendee
3. Generate a 2–3 page brief per meeting, save as @Google Drive docs
4. Send me a @Gmail summary email with links to each brief

Output format: email subject "Tomorrow's Client Meeting Briefs — [Date]", body as table (Client | Meeting Time | Key Topics | Brief Link)

OpenAI internal case: sales team turned Discovery conversations into customized PoC proposals within 24 hours (traditionally weeks).

Scenario B: Account Command Center (Sites + Daily Updates)

Based on all opportunities, contacts, and recent activity for [Account Name] in @Salesforce:

1. Create an interactive account command center (Sites) with: pipeline overview, last-7-day key signals, recommended next actions
2. Set a Scheduled Task: auto-update the Site every weekday at 8:00 AM
3. DM me via @Slack when major changes occur

Constraints: do not auto-send any external emails; amounts must match CRM source data.

Scenario C: Lead Review and Pipeline Repair (Zapier case adapted)

Analyze new leads from the past 30 days in @Salesforce and follow-up records, cross-referenced with @Gmail sales correspondence.

Find: leads not followed up within 48 hours (grouped by source), broken follow-up chains, estimated pipeline loss.

Output: Excel detail sheet + 1-page executive summary PPT (highlighting seven-figure potential loss) + recommendation for a weekly repeatable review workflow.

4.2 Marketing

Scenario A: Research → Brief → Multi-Market Assets (End-to-End)

I uploaded customer research: [attachment / @Google Drive link]

Phase 1 — Brief: extract audience, pain points, competitive positioning; output Campaign Brief (Google Docs)
Phase 2 — Assets: 1 acquisition email, 3 LinkedIn posts, 1 landing page copy outline → save to Drive
Phase 3 — Regional adaptation: US, Europe, APAC versions; flag sensitive phrases needing human review

Pause after each phase and wait for my confirmation before proceeding.

Scenario B: Slack / Teams Activity Sync to Meeting Agenda (Weekly Scheduled)

Set a scheduled task to run every Monday at 7:00 AM:

1. Summarize important discussions from the past 7 days in @Slack #product-launch and @Microsoft Teams "Go-to-Market" channel
2. Extract: decisions made, open questions, blockers
3. Update @Google Drive "Weekly Meeting Agenda" doc (preserve version history)
4. Post a 5-bullet-or-less summary in @Slack #leadership

Constraints: only cite publicly shared discussions; do not leak confidential messages.

4.3 Finance

Scenario A: Month-End Variance Analysis (OpenAI internal validation) — Result: close process compressed from days to hours.

Help complete [Month] month-end budget variance analysis:

Pull spreadsheets from @Google Drive "Finance / Actuals" and "Finance / Forecast"
Create reconciliation workbook in @Google Sheets: actual vs forecast by department, flag items with variance >5% or >$50K
Draft performance narrative (Google Docs), categorized by revenue / COGS / OpEx
Build 5–8 page management presentation PPT (with charts, following attached template style)
List 3 key judgment calls requiring finance human sign-off

Constraints: do not modify source data; cite source cell for every number.

Scenario B: Invoice and Payment Reconciliation (AP Automation)

Compare payment register against invoice list (@Google Drive links).

Flag: amount variance >2%, missing tax IDs, duplicate invoice numbers, vendor name mismatches.
Return review table; do not initiate payments automatically.

4.4 Operations

Scenario A: Daily Dashboard Change Monitoring (Scheduled)

Auto-run every weekday at 6:30 AM:

1. Access [internal dashboard URL / @SharePoint report page]
2. Compare against yesterday's snapshot; extract >10% swings or new red indicators
3. Generate 1-page morning brief (Google Docs): TOP 3 watch items, metric change table, suggested follow-up owners
4. Send via @Gmail to ops-leads@company.com

If dashboard is inaccessible, tell me in Plan phase — do not fabricate data.

Scenario B: Customer Feedback Theme Clustering → Product Priorities

Monitor feedback from the past 14 days: @Slack #customer-feedback, @Gmail label "NPS-Detractor", @Google Drive Support Tickets Export.

1. Cluster into 5–8 themes (with representative quotes)
2. Prioritize by "frequency × impact × implementation difficulty"
3. Output product evaluation backlog; set weekly Friday auto-refresh

Constraints: anonymize feedback quotes; no customer names.

4.5 Product

Scenario A: Cross Jira + GTM Launch Readiness Review (Nvidia case adapted)

Run launch readiness review for [Product/Feature Name]:

1. Pull Epic/Story completion status and open blockers from @Jira
2. Check key milestones from @Google Drive "GTM Plans"
3. Extract unresolved discussions from @Slack #product-launch in the last 7 days
4. Output Readiness report: readiness score (red/yellow/green), blocker list, Go/No-Go recommendation

Do not auto-modify Jira; flag high-risk items requiring human decision.

4.6 Engineering — Work and Codex Collaboration

Engineering workflows work best with Codex for code, Work for cross-team documentation — switch modes in the same desktop app.

Scenario A: PR Review + Release Notes

Codex mode: review [repo/name] PR #123 (security/performance/test coverage), sidebar comments line-by-line; on approval, draft Release Notes.

Work mode: format as @Confluence page, draft @Slack #engineering announcement (do not auto-send).

Scenario B: Multi-Repo Issue Summary Weekly Report

Codex mode: across [frontend-repo] and [backend-repo], summarize this week's merged PRs and open P0/P1 issues; generate engineering weekly report Markdown.

Work mode: convert to Google Docs and insert burndown chart (from @Jira); set Scheduled Task every Friday at 5:00 PM for auto-generation.

5. Scheduled Tasks Automation Recipe Library

Recipe Name	Trigger	Task Description	Best For
Monday Agenda Refresh	Every Monday 07:00	Summarize Slack activity → update agenda doc	Marketing / Ops
Daily Metrics Morning Brief	Every weekday 06:30	Access dashboard → compare yesterday → email brief	Ops / Finance
Feedback Cluster Weekly	Every Friday 16:00	Multi-channel feedback → theme clustering → priority list	Product
Account Activity Daily	Every weekday 08:00	CRM changes → update Sites command center	Sales

5.1 Prompt Pattern for Setting Scheduled Tasks

Set up a Scheduled Task:
- Frequency: [daily / every Monday / 1st of month / when keyword appears in @Slack channel]
- Time: [timezone + specific time]
- Action: [specific workflow description]
- Notification: [Slack channel / email / none]
- Human approval: [which steps require my sign-off first]

5.2 Safety Checklist Before Unattended Runs

Plugin access scope limited (only necessary tools connected)
"Auto-send externally" disabled unless explicitly required
Output archive path set to avoid overwriting others' files
Enterprise: admin-approved Agent network policy confirmed
Run "single execution" 2–3 times to validate before scheduling

6. Usage Optimization: Make Work Mode Cost Less

ChatGPT Work and Codex share a unified usage billing pool. The same workflow designed differently can cost 5× more or less.

Factor	Impact on Usage
Number of task steps	More steps = higher consumption
Context size	More docs/emails pulled = higher consumption
Output length	Output tokens cost roughly 6× input tokens
Cache hits	Re-reading same doc: cached input ≈ 1/10 of fresh input
Model selection	GPT-5.6 complex reasoning costs more than lightweight tasks need

6.1 Seven Cost-Saving Tips

Draft in Chat mode first, then hand a trimmed version to Work
Trim redundant steps in Plan Mode, especially repeated pulls from the same source
Reuse the same template doc in Scheduled Tasks to leverage cache discounts
Keep output concise: "table + 3 bullet summary" beats a full narrative report
Split large tasks: Phase 1 confirm direction → Phase 2 generate deliverable
Free users: run small tasks on desktop first, measure consumption before scaling
Enterprise: set workspace / group / individual quotas in Admin Console

6.2 Five-Step "Usage Estimate" Before Rollout

Pick a real task with known manual time (e.g. month-end variance table, typically 2 hours)
Run once in Work mode with Plan Mode, record step count
After execution, check consumption (compare against plan included usage)
Estimate: if run daily/weekly, is monthly consumption within budget?
If high → optimize per above, re-run and compare

7. Common Pitfalls and Troubleshooting

Problem	Cause	Solution
Work can't find Codex projects	App migration incomplete	Update Codex App → auto-becomes ChatGPT desktop; if broken, reinstall from chatgpt.com/download
Plugin authorized but no data	Insufficient permissions or wrong @name	Check plugin directory authorization; write `@Salesforce` not generic "CRM"
Plan looks right but output drifts	Stale context or AI inference	Pause mid-run to correct; provide key data via attachment/link explicitly
Scheduled task didn't fire	Computer asleep / desktop not logged in	Long-running tasks: use Web Workspace Agent; desktop must stay awake
Usage higher than expected	Output too long, repeated pulls, too many steps	Apply Section 6 optimizations; Enterprise: set limits in Admin Console
Unclear: Work vs Cowork	Different workflow types	Cloud SaaS orchestration → Work; local folder batch processing → Cowork

8. 30-Day Onboarding Roadmap

Phase	Goal	Actions
Week 1	Master single tasks	Pick your most familiar task, run desktop Work manually 3×, practice Plan Mode review
Week 2	Deep plugin integration	Connect 3 core tools (email + collaboration + files), complete 1 cross-app end-to-end delivery
Week 3	Automation	Convert Week 1 task to Scheduled Task, verify 3 successful triggers
Week 4	Team rollout	Build role-specific prompt template library; Enterprise: sync admin on quotas

9. Deep Dive: Virgin Atlantic and OpenAI Internal Finance Validation

OpenAI cited multiple enterprise early-validation cases at the ChatGPT Work launch. Virgin Atlantic compressed sales meeting prep from 1–2 hours of manual work into an automated brief pipeline; OpenAI's internal finance team used Work for month-end variance analysis, cutting close cycles from days to hours. The key isn't "smarter AI" — it's standardizing known repetitive processes into Prompt + Scheduled Task.

For Mac users, these workflows run best in desktop Work mode: Computer Use can drive local Excel, FCP project folders, or Xcode simulators. But when multiple Scheduled Tasks run in parallel (sales briefs + finance close + ops morning brief simultaneously), MacBook unified memory and foreground responsiveness quickly become the bottleneck — which is exactly where a "local Mac for approval + remote Mac for background agents" split architecture pays off.

10. FAQ

Q: Which role workflow should I practice first?

Pick a task you know well enough to judge output quality. OpenAI recommends: month-end variance analysis, marketing briefs, sales meeting prep.

Q: How long should a prompt be?

Focus on "data source + output format + constraints" — typically 150–400 words is enough. Don't write every manual step.

Q: Can scheduled tasks run when my computer is off?

Desktop Scheduled Tasks require the device to be online. For true unattended background runs, use Web Workspace Agent on Plus and above.

Q: What's the difference between Work mode and Workspace Agent?

Work is an individual Agent mode inside ChatGPT. Workspace Agent is a team-built, shared, centrally managed automation Agent within Business/Enterprise, with Admin Console governance.

Q: Can generated PPT/Excel go straight to external stakeholders?

Treat as an "80% first draft." Financial figures, client names, and external statements always need human review.

Q: Which templates can free users run?

Desktop Work is available on trial with usage caps. Start with lightweight tasks like "invoice reconciliation" before long-running automation.

11. References

12. Wrap-Up: Run Work on Mac Desktop, Offload Long Agents to Remote Nodes

ChatGPT Work's value isn't that "it launched" — it's that it eliminates manual processes you've grown tired of. The Mac desktop is the best host for Work + Codex + Computer Use: local Excel reconciliation, folder batch processing, and GUI automation run smoothly on Apple Silicon. But when multiple Scheduled Tasks run in parallel and multi-hour Agent plus Codex multi-repo jobs stack up, unified memory and battery drain fast; pure Windows cloud hosts can run background jobs but offer weak support for Apple ecosystem toolchains and Metal graphics workflows.

A more practical architecture: local Mac handles Work interaction and Plan Mode approval, while sales brief pipelines, finance month-end, ops morning briefs, and other scheduled agents run on MACGPU remote Mac mini M4 nodes — Apple Silicon unified memory suits parallel agents, SSH on-demand start/stop, forming a "foreground control + background resident" two-tier setup with your local ChatGPT desktop.