2026 ChatGPT Work Practical Guide: Role Workflows, Prompt Templates & Scheduled Task Recipes

OpenAI shipped ChatGPT Work on July 9, 2026. The real question is what you run on Monday morning. This guide follows OpenAI's own advice—start with a task you already know—and breaks down copy-paste prompts for Sales, Marketing, Finance, Ops, Product, and Engineering, plus Plan Mode review checklists, Scheduled Task recipes, and usage optimization tactics. For the full launch picture and Cowork comparison, see the companion piece ChatGPT Work launch decision guide.

1. Three pain points: knowing Chat is not knowing Work

1. Pain point one: wrong mode, doubled usage. Running multi-step cross-app jobs in Chat mode—or using Work mode for simple Q&A—burns through the shared metering pool. Mode confusion is the biggest day-one cost trap.
2. Pain point two: prompts written like SOPs. Work plans its own path. If you script "open Salesforce, export, then…" step by step, you constrain the agent and increase rework cycles.
3. Pain point three: unattended automation without a safety review. Auto-sending external email, overwriting shared docs, or pulling the wrong month of data—high-stakes actions need explicit constraints in Plan Mode and Scheduled Task settings.

2. Before you start: three principles that decide success

2.1 Pick the right mode: Chat / Work / Codex quick routing

2.2 Desktop vs web: where to run each workflow

3. Universal five-step workflow and prompt formula

Regardless of role, run every Work job through this sequence:

1. Connect plugins → 2. Define goal and output format → 3. Review Plan Mode → 4. Steer mid-run → 5. Accept deliverables and iterate

3.1 Prompt formula (Work mode)

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

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

3.2 Plan Mode review checklist

• Are data sources correct (wrong customer or wrong month)?
• Does the plan include high-risk actions like external send, delete, or file overwrite?
• Does the output format match your team template?
• Can intermediate steps be trimmed to save usage?
• Do you need a human confirmation gate before certain steps?

4. Sales: three scenario prompts

Templates below draw on OpenAI's published examples and early feedback from Zapier, Nvidia, and Virgin Atlantic pilots. Swap @plugin-name tokens for your stack.

4.1 Scenario A: automated customer meeting brief (daily schedule)

Pain: reps spend 1–2 hours daily assembling account context. Work fix: scan calendar → pull CRM notes → search news → generate brief and archive. OpenAI internal case: discovery calls converted to customized PoC proposals within 24 hours (traditionally weeks).

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

1. Check my tomorrow @Google Calendar customer meetings (exclude internal meetings)
2. For each customer meeting:
   - Pull last-30-day account notes and activity from @SharePoint / @Salesforce
   - Search public news and executive moves 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 and save as @Google Drive docs
4. Send me a @Gmail summary email with links to each brief

Output format: email subject "Tomorrow's Customer Meeting Briefs — [date]"; body is a table (Customer | Meeting time | Key topics | Brief link)

4.2 Scenario B: account command center (Sites + daily refresh)

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

1. Create an interactive account command center (Sites) with:
   - Pipeline overview (stage, amount, expected close date)
   - Last-7-day key signals (email, meetings, support tickets)
   - Recommended next actions ranked by priority
2. Set a Scheduled Task to refresh this Site every weekday at 8:00 AM
3. DM me on @Slack when material changes occur

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

4.3 Scenario C: lead review and pipeline repair (Zapier case adapted)

Analyze new leads from the last 30 days in @Salesforce and their follow-up history; cross-reference sales correspondence in @Gmail.

Find:
1. Leads with no follow-up within 48 hours (grouped by source)
2. Broken follow-up chains (where response rate drops sharply after a step)
3. Estimated pipeline loss in dollars

Output:
- Excel detail sheet (Lead ID | Source | Last follow-up | Break type | Recommended action)
- One-page executive summary slide deck highlighting seven-figure opportunity risk
- A weekly repeatable review workflow suitable for Scheduled Tasks

5. Marketing: two scenario prompts

5.1 Scenario A: research → brief → multi-market assets (end-to-end pipeline)

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

Run an end-to-end marketing workflow:

Phase 1 — Brief:
- Extract target audience, core pain points, competitive positioning
- Output a Campaign Brief (Google Docs) with message pillars and channel recommendations

Phase 2 — Asset generation:
- From the brief, produce: one acquisition email, three LinkedIn posts, one landing-page copy outline
- Save to @Google Drive folder "Campaign / [Product Name]"

Phase 3 — Regional adaptation:
- Adapt core assets for US, Europe, and APAC (language, cultural references, compliance wording)
- Flag sensitive phrases requiring human review in each version

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

5.2 Scenario B: Slack / Teams activity into meeting agenda (Scheduled Task)

Set a scheduled task for every Monday at 7:00 AM:

1. Summarize important discussions from the last 7 days in @Slack #product-launch and @Microsoft Teams "Go-to-Market"
2. Extract: decisions made, open questions, blockers needing alignment
3. Update the "Weekly Meeting Agenda" doc on @Google Drive (preserve version history)
4. Post a summary of five bullets or fewer in @Slack #leadership

Constraints: quote only publicly discussed content; do not leak messages marked confidential.

6. Finance: two scenario prompts

6.1 Scenario A: month-end variance analysis (OpenAI internal validated scenario)

OpenAI internal result: month-end close and forecast workflows compressed from days to hours.

Help complete [month] budget variance analysis:

1. Pull corresponding tables from @Google Drive "Finance / Actuals" and "Finance / Forecast"
2. Create a reconciliation workbook in @Google Sheets:
   - Summarize actual vs forecast variance by department
   - Flag line items with variance >5% or >$50K
   - Preserve all source formulas; do not overwrite source files
3. Draft performance narrative (Google Docs) explaining likely drivers by Revenue / COGS / OpEx
4. Build a 5–8 slide management deck (charts, following attached template style)
5. List three key judgments requiring finance human sign-off

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

6.2 Scenario B: invoice and payment reconciliation (AP automation entry point)

You are an accounts payable specialist. Compare:
- Payment register: [@Google Drive link]
- Invoice list: [@Google Drive link]

Flag anomalies (return as a table):
| Issue type | Vendor | Invoice # | Amount | Recommended action |
- Amount variance >2%
- Missing tax ID
- Duplicate invoice number
- Vendor name mismatch

Do not initiate payments; output review sheet for human approval only.

7. Operations: two scenario prompts

7.1 Scenario A: daily dashboard change monitoring (Scheduled Task)

Run automatically every weekday at 6:30 AM:

1. Access [internal dashboard URL / @SharePoint report page]
2. Compare to yesterday's snapshot; extract significant changes (>10% swing or new red indicators)
3. Generate a one-page morning brief (Google Docs):
   - TOP 3 items needing attention today
   - Metric change table
   - Suggested owner for follow-up
4. Email ops-leads@company.com via @Gmail

If the dashboard is unreachable, tell me in Plan Mode—do not fabricate data.

7.2 Scenario B: customer feedback clustering → product priority

Monitor new customer feedback from the last 14 days across:
- @Slack #customer-feedback
- @Gmail label "NPS-Detractor"
- @Google Drive "Support Tickets Export"

1. Cluster feedback into 5–8 themes (with representative quotes)
2. Score priority by Frequency × Impact × Implementation difficulty
3. Output a product evaluation backlog (Notion / Google Docs format)
4. Set a Scheduled Task to refresh this doc every Friday

Constraints: anonymize quotes; no customer names in output.

8. Product: launch readiness review (Nvidia case adapted)

Run a launch readiness review for [product/feature name]:

1. Pull linked Epic / Story completion and open blockers from @Jira
2. Pull the corresponding GTM plan from @Google Drive "GTM Plans" and check milestone status
3. Extract unresolved discussions from @Slack #product-launch in the last 7 days
4. Output a Launch Readiness report (Google Docs):
   - Readiness score (Red / Yellow / Green)
   - Blocker list (Owner | Due date | Risk level)
   - Recommended Go / No-Go with rationale

Do not auto-update Jira status; flag high-risk items for human decision.

9. Engineering: Work and Codex together

Engineering teams should let Codex own code and Work own cross-team documents—switch modes inside the same desktop app without changing tools.

9.1 Scenario A: PR review + release notes (Codex-led)

In Codex mode:
1. Review PR #123 in [repo/name], focusing on [security / performance / test coverage]
2. Leave line-by-line review comments in the PR sidebar
3. If approved, draft Release Notes

Switch to Work mode:
4. Format Release Notes as a @Confluence page
5. Draft an announcement for @Slack #engineering (do not auto-send)

9.2 Scenario B: multi-repo issue rollup weekly report (Codex multi-repo)

In Codex mode, across [frontend-repo] and [backend-repo]:
1. Summarize merged PRs this week and open P0/P1 issues
2. Generate an engineering weekly report in Markdown

Switch to Work mode:
3. Convert to Google Docs and insert this week's burndown chart (from @Jira)
4. Set a Scheduled Task for every Friday at 5:00 PM to auto-generate

10. Scheduled Tasks recipe library

10.1 Scheduled Task setup prompt template

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

10.2 Safety checklist before unattended runs

• Plugin access scoped to necessary tools only
• Auto external send disabled unless explicitly required
• Output archive path set to avoid overwriting shared files
• Enterprise: admin-approved agent network policy confirmed
• Run manually 2–3 times successfully before enabling schedule

11. Usage optimization: spend less on Work

ChatGPT Work and Codex share a single usage pool. The same workflow, designed differently, can cost up to 5× more.

11.1 Billing logic (simplified)

11.2 Seven cost-saving tactics

1. Draft in Chat mode first, then hand a trimmed prompt to Work
2. Trim Plan Mode steps, especially duplicate pulls from the same source
3. Reuse the same template doc in Scheduled Tasks to benefit from cache discounts
4. Demand concise output: "table + three bullets" beats a narrative report
5. Split large jobs: Phase 1 confirms direction → Phase 2 generates deliverables
6. Free tier: run small desktop tasks first; measure before scaling
7. Enterprise teams: set workspace / group / per-user caps in Admin Console

11.3 Five-step usage estimate before production

1. Pick a real task you already know the manual time for (e.g., month-end variance table, usually 2 hours by hand)
2. Run once in Work with Plan Mode and record step count
3. Check consumption against your plan's included usage
4. Estimate monthly cost if run daily or weekly
5. If high → optimize per section 11.2 and re-run to compare

12. Common pitfalls and troubleshooting

13. 30-day adoption roadmap

14. FAQ

Which role workflow should I practice first?
Pick a task you already know well enough to judge output quality. OpenAI recommends month-end variance analysis, marketing briefs, and sales meeting prep.

How long should a Work-mode prompt be?
Focus on data sources, output format, and constraints—150 to 400 words is usually enough. Do not write step-by-step manual instructions.

Can Scheduled Tasks run while my laptop is off?
Desktop Scheduled Tasks require the device online. For true unattended background runs, use web Workspace Agent on Plus or above.

What is the difference between Work mode and Workspace Agent?
Work mode is personal agent mode inside ChatGPT. Workspace Agent is team-built, shared, admin-governed automation in Business and Enterprise.

Can I ship AI-generated PPT or Excel files externally?
Treat outputs as 80-percent drafts. Finance figures, customer names, and external statements need human review.

Which templates can free-tier users run?
Desktop Work is available with usage caps. Start with lightweight tasks like invoice reconciliation before long automation.

Sources: OpenAI official blog, Sales Meeting Prep Cookbook, ChatGPT Learn Changelog, SiliconANGLE. Last updated: 2026-07-11.

15. Summary: run Scheduled Tasks on an always-on Mac

ChatGPT Work is not "another AI feature"—it is built to finish the manual workflows you already resent. Fastest ROI: pick a task you know → run manually three times to tune the prompt → build trust in Plan Mode → let Scheduled Tasks handle repetition. The twelve role prompts and four scheduled recipes in this guide are ready to copy and adapt.

Desktop Scheduled Tasks and Computer Use share one hard constraint: the execution host must stay online, awake, and network-stable. Laptops that sleep or home Wi-Fi that drops will kill a 6:30 AM metrics brief or 8:00 AM account refresh mid-run—regardless of prompt quality. That is a deployment bottleneck, not a model problem.

If you plan production-grade scheduled agents (sales briefs, month-end reconciliation, engineering weekly reports), anchor ChatGPT desktop on an always-on Apple Silicon Mac and sync workspaces via SFTP or rsync. SFTPMAC remote Mac rental targets AI Agent workloads: native Apple ecosystem compatibility, low-latency plugin callbacks, and 24/7 uptime—a better production host than a personal laptop you close every evening.