What are AI agents?
AI agents, under Create → Automations, are goal-driven workers that pick their own tools, branch and reason between steps to research, write and optimise content.
AI agents live under Create → Automations. They are goal-driven workers rather than fixed, linear pipelines: you give an agent a goal and a set of tools, and it decides which tools to use, and in what order, to reach that goal. This article explains how they work and how to build, run and schedule one for your brand.
When an agent runs, it plans first. The AI model you chose then picks which of its granted tools to call, reasons over the results and carries on until the goal is met or it reaches one of its guardrails. Because the agent chooses its own path, it can branch on a condition and adapt as it goes, which suits research, writing and optimisation tasks that are different every time. Every agent always reads your brand data from Context, so its output stays on brand.
Tools are grouped on the canvas
- Research & SEO — Web search, Google search, Scrape page, Deep research, Keyword research, SERP analysis, Gap analysis and more.
- Content — Outline, Title suggestions, Generate article, Create brief, FAQ cluster, Internal linking plan and Social pack.
- Optimisation — Editorial review, Content scorecard, Remove AI patterns, SEO format and Schema markup.
- AI visibility and WriteWorks Data — your visibility score, share of voice, citations, prompt volumes, content scores and opportunities.
- Brand & Assets, Knowledge Base and Logic — Use persona, Use skill, Use brand kit, Search knowledge base, LLM reasoning and Branch / decision.
- Human review — a Review point pauses the run for a person to approve. Tools that write content are marked Requires approval.
How to build and run an agent
- Go to Create → Automations.
- Choose Build with AI to describe what you want in a chat and let WriteWorks build the agent, Build manually to work on the canvas, or Browse templates to start from one such as AI visibility audit, Competitor AI watch, New article or Keyword → brief pipeline.
- In the agent's settings, set the Goal, choose its Tools, decide whether to review the plan before it runs, and set the guardrails: Max steps and Token budget.
- Pick the AI model from the model picker. Each model shows its usage rate compared with the default.
- Choose where the result goes: Open in editor, Email the result, Notify me in the app, File to a project board, Add a row to a Sheet, Post to Slack, Send to a webhook or Download.
- Save, then run it now or schedule it hourly, daily, weekly or monthly.
- My automations lists everything you have saved, and Scheduled shows upcoming runs.
- Run history records every run with its status, outputs and success rate.
- Runs use your organisation's AI tokens, weighted by the model you choose.
Tip: if you are new to agents, start from a template, adjust the goal and tools, and keep a Review point before anything is published.
Still stuck? We typically reply within 1 business day.
Contact support