Inside the Codex section of the ChatGPT desktop app, sets up the Zapier SDK, links your social accounts or pages, fixes a failed connection from a screenshot, then writes and publishes a post, with an optional matching image, to the page you linked.
The task I repeat: [describe the task] Task: turn it into a simple plan for automating it. The plan shows: - the trigger - each action, in order - the tools required - checks that catch failures - which steps still need approval from a person
Role: an automation architect. Tool: [your AI agent, e.g. ChatGPT 6 Astra] together with the business tools it can connect to. Business: [describe the business] Task: 1. Analyze this business and spot the repetitive tasks that the tool above could automate. 2. Pick out the automation opportunities with the highest value, covering customer onboarding, lead tracking, follow-ups, data entry, reporting and any other process that repeats. 3. For each automation, define: - trigger - actions - tools required - workflow logic - how to test it - what maintenance it needs 4. Then design the simplest workflow that is still reliable, one I can build for this business and charge for as a one-off setup plus a recurring support fee.
Role: an expert in optimizing workflows. Task or process: [your task or process] How I do it today: [describe your current steps] Task: 1. Analyze my current way of doing it. 2. Find the steps that repeat, the bottlenecks and the places where AI or simple tools could automate the work. 3. Design a faster, more efficient workflow and give me clear steps to implement it.
Role: you are a top AI automation consultant who also thinks like an operations strategist, a workflow architect and a productivity expert. My business: [describe your business] My daily tasks: [list them] Tools I use: [list them] Current bottlenecks: [describe them] Task: 1. Analyze the information above. If something essential is missing, ask me before continuing. 2. Design a complete operating system for AI automation that cuts repetitive work while raising quality and reliability. 3. Sort every task into one of three groups: fully automatable, AI-assisted, human-only. 4. For each recommendation, explain why you made it. Present the task classification as a table.
Role: you are a process optimization consultant for enterprises. My business and its recurring workflows: [describe them] Task: audit each recurring workflow and find where it can be automated. For each workflow, estimate: - time saved - difficulty of implementation - impact on the business - risks - AI or automation tools you suggest Order the automations so the ones with the highest leverage come first. Present the result as a table. Constraints: only automations that are practical to implement, aimed at productivity gains you can measure.
Role: you are an architect of AI systems. My business: [describe your business] Task: design a virtual team of AI roles for this business, each one specialized. Examples of roles to assign: - assistant to the CEO - researcher - copywriter - marketer - agent for customer support - data analyst - software engineer - financial analyst - operations manager For each role, define: - responsibilities - inputs - outputs - how its work is reviewed - when a human needs to step in Constraints: design for AI and people working together, not for AI replacing people entirely.
Role: you are a marketing director specialized in AI. My business: [describe your business] Task: build automated systems for each of these marketing activities: - market research - content planning - drafting copy - email marketing - SEO support - scheduling social media posts - analytics reporting - optimizing campaigns For each output, state clearly whether it must be reviewed before it is published. Constraints: marketing that puts the customer first; aim for efficiency and quality that stays consistent.
Role: you are a consultant for AI in sales operations. My business: [describe your business] Task: design AI-assisted systems for: - qualifying leads - updating the CRM - drafting outreach - onboarding - managing the knowledge base - triaging support tickets - generating FAQs - reporting on customer success Point out where human judgment adds the most value. Constraints: put customer experience, transparency and accuracy first.
Niche: [niche] Task: design the logic of an AI bot for customer service in this niche. Build it as a decision tree that covers: - refunds - FAQs - product tracking The bot must resolve all of these without a human ever having to step in.