How to use ChatGPT Work: 3 principles before you copy a prompt
| Principle | What it means | Practical tip |
|---|---|---|
| Describe outcomes, not steps | Work mode plans its own path | Not "Open Salesforce, export…" — instead "Build a weekly pipeline PPT from @Salesforce deals in the last 30 days, flagging at-risk opportunities" |
| Connect tools first | Plugins are Work's data layer | Authorize Gmail, Slack, Drive before starting; use @AppName to pin sources |
| Plan Mode is your brake | Review the plan before execution | For high-stakes deliverables (external emails, financial reports, client docs), approve every step |
| Your need | Use | Why |
|---|---|---|
| Quick Q&A, brainstorming, single-turn copy | Chat | Lightweight, fast |
| Multi-app projects, finished deliverables, hours-long tasks | Work | Plugins + Plan Mode + Computer Use |
| Code review, PRs, multi-repo development | Codex | Developer-native workflows |
| Recurring background automation | Work + Scheduled Tasks | Triggered or scheduled execution |
Local files & Computer Use: Desktop app (Mac/Windows) for local file access and free-tier trials.
Team visibility: Web/mobile (Plus+) to monitor task progress anywhere.
Sales meeting briefs: Web Workspace Agent + scheduling for email notifications.
Local Excel reconciliation: Desktop Work mode with Computer Use.
Core pain point: Regular Chat gives advice; Work pulls data across tools, runs the workflow, and delivers usable files — if you specify the deliverable, not the clicks.
ChatGPT Work prompt formula and 5-step workflow
1. Connect plugins → 2. Write goal + output format → 3. Review Plan Mode → 4. Steer mid-flight → 5. Accept deliverable & iterate
[Role] + [Data sources @plugins] + [Task] + [Output format] + [Constraints] + [Acceptance criteria] You are [role]. Pull [data type] from @Salesforce and @Gmail for [time range]. Complete [action], output as [Google Docs / Excel / PPT / Sites]. Constraints: [do not modify source data / 2 decimal places / no external email]. Then [Slack notify me / save to folder].
Data sources: Right account, right month?
High-risk actions: Send external email, delete, or overwrite files?
Output format: Matches your team template?
Step trimming: Can duplicate data pulls be removed to save usage?
Approval checkpoints: Which steps need your sign-off?
Dry runs: Run 2–3 manual tests before scheduling high-stakes automation.
6 role-based ChatGPT Work workflows with prompt templates
Templates draw from OpenAI case studies, early adopters (Zapier, Nvidia, Virgin Atlantic), and the Workspace Agent Cookbook. Swap @plugin names for your stack.
Sales
A. Daily meeting briefs (scheduled): Scan tomorrow's calendar, pull CRM notes, search news, archive briefs. OpenAI internal case: Discovery call to custom PoC in 24 hours (weeks traditionally).
Create a scheduled task running every weekday at 4pm: 1. Check tomorrow's customer meetings in @Google Calendar (exclude internal-only) 2. For each: pull 30-day account notes from @Salesforce, search 30-day company news 3. Write a 2–3 page brief per meeting, save to @Google Drive 4. Email me a summary with links via @Gmail
B. Live account command center (Sites + daily refresh): Interactive dashboard from @Salesforce with pipeline, 7-day signals, prioritized actions. Daily 8am refresh; Slack on major changes. No external emails.
C. Lead review & pipeline repair (Zapier-style): Cross-reference @Salesforce leads with @Gmail outreach. Find 48h+ follow-up gaps, broken handoffs, estimated pipeline loss. Output Excel detail + executive PPT + repeatable weekly review workflow.
Marketing
A. Research → Brief → Multi-market assets: Phase 1 Campaign Brief; Phase 2 email, LinkedIn posts, landing page outline; Phase 3 US/EU/APAC localization with sensitive-term flags. Pause after each phase.
B. Slack/Teams → meeting agenda (weekly scheduled): Monday 7am: summarize #product-launch and GTM channel; update Weekly Agenda doc; post ≤5 bullets to #leadership. Public content only.
Finance
A. Month-end variance analysis (OpenAI-validated): Pull actuals/forecast from Drive, build reconciliation in Sheets (flag >5% or >$50K), draft narrative, 5–8 slide deck, list 3 human judgment calls. Internal result: days → hours.
B. Invoice vs. payment register: Flag >2% amount differences, missing tax IDs, duplicate invoice numbers, vendor mismatches. Review table only — no payments initiated.
Operations
A. Daily dashboard briefing (scheduled): Weekdays 6:30am — compare dashboard to yesterday (>10% swings), 1-page brief to ops-leads. Stop and notify if unreachable — never fabricate data.
B. Feedback clustering → product priorities: 14-day feedback from Slack, Gmail NPS-Detractor, Drive tickets. Cluster 5–8 themes, rank by frequency × impact × effort. Anonymize references; schedule Friday refresh.
Product
A. Launch readiness (Jira + GTM, Nvidia-style): Jira completion/blockers, GTM milestones, Slack unresolved threads. Red/Yellow/Green report with Go/No-Go. Do not auto-update Jira.
Engineering — Work + Codex in one app
| Scenario | Codex mode | Work mode |
|---|---|---|
| A: PR review → release notes | Review PR #123, side-panel comments, draft release notes | Format for @Confluence, draft @Slack #engineering post (do not auto-send) |
| B: Multi-repo weekly summary | Cross frontend + backend — merged PRs, open P0/P1 issues → Markdown | Convert to Google Docs, Jira burndown; schedule Fridays 5pm |
ChatGPT Work Scheduled Tasks recipes and six-step runbook
| Recipe | Trigger | Action | Best for |
|---|---|---|---|
| Monday agenda refresh | Mon 7am | Slack digest → update agenda doc | Marketing / Ops |
| Daily metrics brief | Weekdays 6:30am | Dashboard diff → email report | Ops / Finance |
| Feedback clustering | Fri 4pm | Multi-channel → priority list | Product |
| Account daily refresh | Weekdays 8am | CRM changes → update Sites dashboard | Sales |
Set Scheduled Task: - Frequency: [daily / every Monday / 1st of month / Slack keyword trigger] - Time: [timezone + exact time] - Action: [workflow description] - Notify: [Slack channel / email / none] - Human approval: [steps requiring my sign-off]
Download desktop app from chatgpt.com/download; update existing Codex install.
Switch to Work mode in the top navigation.
Connect tools: Authorize Gmail, Slack, Drive, Salesforce in the plugin directory.
Pick a role template from Section 03; replace @plugin names and date ranges.
Review Plan Mode for data sources, risky actions, and output format.
Accept and automate: Configure Scheduled Tasks; use web Workspace Agents for unattended long cycles.
ChatGPT Work usage optimization, pitfalls, and 30-day roadmap
ChatGPT Work shares a metered usage pool with Codex. The same workflow can cost 5× more depending on design.
| Factor | Impact on usage |
|---|---|
| Step count | More steps = more consumption |
| Context size | Larger document/email pulls cost more |
| Output length | Output tokens cost ~6× input |
| Cache hits | Repeated reads: cached input ~1/10 of fresh |
| Model choice | GPT-5.6 heavy reasoning exceeds light-task needs |
Draft in Chat first, then hand a tight brief to Work.
Trim Plan Mode steps, especially duplicate data pulls.
Reuse template docs in Scheduled Tasks for cache discounts.
Request concise outputs: table + 3 bullets beats narrative reports.
Split large projects into phases to avoid expensive re-runs.
Enterprise: Set workspace/group/individual limits in Admin Console.
| Issue | Cause | Fix |
|---|---|---|
| Codex projects missing | Incomplete app migration | Update Codex app → becomes ChatGPT desktop; clean reinstall if broken |
| Plugin connected but no data | Insufficient scope or wrong @name | Re-check permissions; use explicit @Salesforce |
| Good plan, wrong output | Stale context or AI inference | Pause and steer; attach explicit source files |
| Scheduled task didn't fire | Device asleep / logged out | Use web Workspace Agents for true background |
| Work vs. Cowork confusion | Different workflow types | Cloud SaaS → Work; local folder batch → Cowork |
| Week | Goal | Action |
|---|---|---|
| 1 | Single-task fluency | Run 3 manual Work tasks you can quality-check |
| 2 | Plugin depth | Connect 3 core tools; complete 1 cross-app deliverable |
| 3 | Automation | Convert Week 1 task to Scheduled Task; verify 3 triggers |
| 4 | Team rollout | Document role-specific prompt library; set admin limits (Enterprise) |
Month-end compression: OpenAI validated closing from days to hours.
5× cost spread: Workflow design can multiply usage cost by five; cached input ~1/10 of fresh.
Output token premium: Output costs ~6× input — favor tables and bullet summaries.
ChatGPT Work pays off when it removes a workflow you already resent doing manually. Desktop Scheduled Tasks need the device awake; laptops sleep, contend for RAM, and drop network. For production environments that need 7×24 Work/Codex runs with stable local file and IDE access, MESHLAUNCH Mac Mini cloud rental is usually the better fit: dedicated Apple Silicon, always-on, flexible daily/weekly/monthly billing as a dedicated ChatGPT Work execution node.
The task you know best and can verify — month-end variance, campaign brief, or sales meeting prep. See our pricing page for dedicated execution nodes.
150–400 words focusing on data sources, output format, and constraints. Do not micromanage steps — that is what Work mode automates.
Desktop tasks need the device online. For true background automation, use web Workspace Agents (Plus+).
Work is personal agent mode inside ChatGPT. Workspace Agents are team-built, admin-governed automations in Business/Enterprise with Admin Console governance.
Treat them as 80% drafts. Always human-review numbers, names, and external statements.
Desktop Work with limits. Start with lightweight tasks like invoice reconciliation before scheduling automation. See our help center for cloud Mac Agent deployment.