01 The AI already in your work
AI did not arrive in comms as a single visible event. It arrived as a feature update inside tools your team was already running. Microsoft Copilot quietly switched on inside Word. Otter or Read.ai joined a meeting without anyone announcing it. Canva added Magic Studio to the toolbar. Streem or Meltwater started summarising coverage with AI. By the time anyone got around to writing a policy, the tools were embedded.
This is not a problem to fix retroactively. It is the starting point to be honest about. You are not deciding whether to start using AI. You are deciding whether to start using it on purpose.
The first step is naming what is already happening. The table below is the structure for that audit. You can use it now or come back to it after you read about the five categories. You will fill it in your AI Tools Shortlist template.
The audit table
| AI element | How it is used | Content type | Generation | Distribution |
|---|---|---|---|---|
| Example: Microsoft Copilot in 365 | Drafting emails, summarising documents, meeting recaps | Internal text, meeting notes, briefings | AI drafts first pass, I edit before sending | Internal: emails, Teams chats, OneNote |
How to use each column
- AI element. The name of the tool or feature. Be specific (Microsoft Copilot in Outlook, not just Copilot).
- How it is used. The actual job the tool does in your work. Drafting, summarising, transcribing, image generation, monitoring.
- Content type. What the AI produces. Email drafts, meeting notes, social posts, briefings, infographics, audio.
- Generation. The role AI plays in producing the content. A first pass you edit? A polish of your draft? The final version unchanged?
- Distribution. Where the content goes. Internal channels only (Teams, email, intranet) or external (media, public, regulators, donors).
The distinction between generation and distribution matters more than it might seem. AI used internally with human review before external publication is a different risk picture from AI used to produce content that goes straight to the public. The audit makes that distinction visible.
Open your Shortlist template and fill in one row for each AI tool or feature currently in your work. Use the example row as the model. Do not filter. Do not judge. Just list. If you are not sure whether something counts as AI, list it anyway and mark it with a question mark. You can resolve it later.
Communications practitioners typically underestimate the AI already embedded in their work by a meaningful margin. Asked to list AI tools in use, most named two or three. Asked to audit their actual workflows across writing, design, meeting, monitoring, and email tools, the same practitioners identified between six and twelve AI elements already in active use. The gap matters because what is unnamed cannot be governed. This audit is the first defence against that gap.
Most communicators list six to twelve items on the first thorough pass. If you list two, you may be missing some: check inside Microsoft 365 or Google Workspace, your CMS, your monitoring platform, your design tools, and your phone. If you list twenty, you may be listing features rather than tools, so focus on the distinct platforms. If you list nothing, that is also a valid result. It tells you something important about your starting point.