V – The Four Agent Jobs

The best tasks for agents are repetitive, time‑consuming, error‑prone or scalable.

At the time of writing, I admit there are probably some major agent jobs that I am missing for non technical creators. But this is what my workflow looks like and most other creators I've spoken to.

For non‑technical creators, those tasks fall into four categories:

  1. Research & Understanding

    1. Use Deep Research agents (OpenAI, Gemini, Perplexity or Claude) when you need a proper report with citations. They'll scan the web, compare sources and synthesise findings at senior analyst level
    2. For quicker web tasks, Browser agents like Perplexity Comet can gather information, compare products and summarise pages in seconds
    3. Autonomous task agents like Genspark are ideal for multi-step research: scanning papers, identifying trends and even producing slides or spreadsheets from what they find

    <aside> 🤖

    Need help writing an AI agent prompt? See the Research Agent Prompt Template for a full example you can copy and adapt

    </aside>

  2. Content & Communication

    1. Let No-code agent builders (Relevance AI, Lindy, n8n) draft emails, briefs, blog posts and social updates on autopilot
    2. Workspace agents like Notion AI can turn meeting notes into action items or write launch plans using your existing docs as context
    3. For end-to-end drafting with tool use, Personal action agents (ChatGPT's Agent mode) can click, scroll, download files and chain tools together to finish a content task from start to finish

  3. Planning & Organisation

    1. Agents can turn messy notes into project plans, roadmaps and checklists. A Workspace agent like Notion AI can break a project into tasks and assign them
    2. While an Autonomous task agent like Genspark can generate slides, spreadsheets and structured plans from your raw ideas

  4. Documentation & Knowledge

    This use case is adaptable. The right combination depends on your own workflow

    1. Use agents to create SOPs, FAQs and internal guides. They can summarise long documents or turn transcripts into actionable insights
    2. Personally, I use a mix of Deep Research agents and Browser agents (for summarising and synthesising), Workspace agents (for organising and storing knowledge) and Personal action agents (for pulling information together end-to-end)

You might find a different combination works better for you — experiment and see what sticks.

A simple rule: if a task is repetitive, text‑heavy and rules‑based, delegate it.