Why "Which Tool?" Comes First
In Level 1 you saw what AI can and cannot do in agency recruitment. Before you use it on live work, you need to answer a simpler question: which tools and features are you actually allowed to use?
Think of it like the difference between driving a pool car and driving your own car on company business. It is the same road, but the insurance, the rules and who carries the risk are different. AI assistants work the same way. ChatGPT, Gemini, Claude and Copilot each come in consumer versions (personal accounts) and business or enterprise versions (managed by your employer), and they handle your data very differently.
This lesson has three tasks. Do each one as you read.
Task 1: Find and Summarise Your Agency's AI Policy
DO: Track down the policy and reduce it to five lines.
- Search internal sources. Look in the staff handbook, intranet or shared drive for: "AI", "generative AI", "acceptable use", "IT security" and "data protection". AI rules are often part of a wider IT or data policy.
- Ask the right people. If you can't find it, ask your line manager, compliance lead, IT team or Data Protection Officer.
- Write a five-line summary:
- Approved tools, including the exact plan (for example "Microsoft 365 Copilot through our work login")
- Approved accounts (work account only, or personal accounts allowed)
- Tasks allowed (for example drafting adverts, summarising your own notes)
- Tasks or data that are banned
- Who to ask, and how to request a new tool
If there is no written policy, don't treat that as permission. Send your manager these questions in writing:
- Which AI tools and plans are approved for work use?
- Can I use a personal account, or only a company-managed one?
- Which tasks are allowed, and which are off-limits?
- Are AI features inside our CRM, email or browser covered?
- Who decides on new tools, and how do I request one?
Task 2: Compare Consumer and Business Versions
Compare any assistant on three points:
- Training: Are your inputs used to train the vendor's models?
- Retention: How long are conversations kept, and can people at the vendor read them?
- Admin controls: Can your employer manage settings, retention and audit?
Here is what the vendors' own pages said in October 2026. Always check the current page yourself, because these terms change.
| Assistant | Consumer version | Business / enterprise version |
|---|---|---|
| ChatGPT | Free, Plus and Pro: data sharing for training is on by default. You can switch it off under Settings > Data Controls > "Improve the model for everyone". | ChatGPT Business and Enterprise: inputs and outputs are not used for training by default. |
| Gemini | With Keep Activity on, chats are used to improve Google services, including training AI models. Some chats are read by human reviewers, and reviewed chats can be kept for up to three years. With Keep Activity off, chats are still kept for 72 hours. | Gemini in a qualifying Google Workspace edition: content is not reviewed by humans or used for model training outside your organisation without permission. |
| Claude | Free, Pro and Max: since September 2025, users choose whether their chats are used for training. If you allow it, data is kept for up to five years. If you don't, the 30-day retention still applies. | Claude for Work (Team and Enterprise) is not covered by those consumer training terms. |
| Microsoft 365 Copilot | Personal Copilot accounts have separate consumer terms. | With enterprise data protection, prompts and responses are not used to train foundation models. They are logged and kept for audit and eDiscovery, and follow your organisation's admin settings. |
The key lesson: a tool's brand name tells you very little. What matters is the plan and the account you're signed into. Gemini on your personal Gmail and Gemini through your agency's Workspace login are covered by different data terms.
DO: Choose the assistant you use most. Open the vendor's current help or privacy page. Note the answer on each of the three points and the date you checked.
Task 3: Decide Whether a Tool or Feature Is Approved
You'll keep running into new AI features: a "summarise candidate" button in your CRM, a browser extension that writes LinkedIn messages, a meeting note-taker. Run each one through these steps:
- Name it exactly: the product, the plan, and whether you're on a personal or work login.
- Check the approved list. Is this exact tool and plan on your policy summary?
- Check the task. Is this use allowed? A tool can be approved for writing adverts but not for handling candidate records.
- Treat built-in features separately. An approved CRM does not automatically mean its new AI add-on is approved. The feature may send data to a different provider. Plugins and browser extensions may read the page you're viewing.
- If any answer is "not sure", stop and ask. Use an approved alternative while you wait.
- Record the decision: who confirmed it, when, and for which tasks.
Example: Priya, a resourcer, sees a new "AI shortlist" button in her CRM. The CRM is approved, but the policy doesn't mention the AI feature. She doesn't use it. She emails compliance asking whether it's covered and which provider processes the data, and sticks to her normal process until she gets a written answer.
Your Task: Try It on Your Own Work
- Write your five-line policy summary, or send the questions above.
- Fill in the three comparison points for one assistant, with the date you checked.
- Pick one AI feature you've noticed in your CRM, email or browser and run it through the six steps.
Checklist
- I know where my agency's AI policy is, or I've asked for one in writing
- I know which plan and account I'm approved to use
- I've checked training, retention and admin controls on the vendor's current page
- I treat CRM add-ons, plugins and extensions as separate decisions
- I stop and ask when I'm unsure
Key Takeaways
- No written policy does not mean permission. Ask in writing.
- Consumer and business versions of the same assistant handle data differently. Check the plan and the account, not just the brand.
- Compare tools on three points: training, retention and admin controls, using the vendor's current help pages.
- An approved platform does not mean its new AI feature is approved.
- Record who approved what, when, and for which tasks.