Meridian Advisory Group
AI tool register & risk summary
Every AI tool reported across Meridian Advisory Group so far, risk-ranked in plain language — not a network scan, a structured self-report your team builds together.
High risk
3
Medium risk
1
Low risk
2
Total tracked
6
How the risk rating works: rules, not a black box
A rating is a lookup on two answers: whose account the tool runs on, and how sensitive the data it touches is. The same answers always give the same rating, and nothing else changes it: not the vendor, not the tool's name. Every rating shows its reason and a recommended action.
| Account type | No sensitive data | Internal | Confidential | Regulated |
|---|---|---|---|---|
| Personal account | medium | high | high | high |
| Not confirmed | medium | medium | high | high |
| Business tier | low | low | low | medium |
- Business tier: a company-managed account under your own agreement with the vendor.
- Personal account: an individual's own account, so the vendor's terms are theirs, not the company's.
- Not confirmed: nobody has checked yet. It is rated no lower than the matching business tier and no higher than a personal account.
Tools like hiring screeners and call recorders also get a “Regulated context” note. It is a pointer, never part of the rating. This is a readiness aid that shows where to look first; it is not a compliance assessment or a certification.
Report a new tool — the form is below the register. Bring your own tools — paste or upload your team's list and see it rated by these same rules.
AI tools on the register (6)
- high
ChatGPT
Sales
- Account
- Personal account
- Data
- Confidential
- Why this rating
- Personal-tier account handling confidential data — the data-handling terms your business agreed to do not apply here.
- Recommended action
- Stop using it with client or confidential data now and move the work to a company-managed account.
- Owner
- Head of Sales
- Notes
- Used to draft client emails and proposals.
ApprovalNot approved - low
Claude
Engineering
- Account
- Business tier
- Data
- Confidential
- Why this rating
- Business-tier account: the company's own agreement with the vendor covers the confidential data it handles. Stays low while the account tier and data scope stay as reported.
- Recommended action
- Keep it on the business account and confirm the vendor's data-handling and retention terms at the next quarterly review.
- Owner
- Engineering lead
- Notes
- Code review and architecture notes.
ApprovalApproved - high
AI note-taker (browser ext.)
Customer Success
- Account
- Personal account
- Data
- Confidential
- Why this rating
- Personal-tier account handling confidential data — the data-handling terms your business agreed to do not apply here.
- Recommended action
- Stop using it with client or confidential data now and move the work to a company-managed account.
- Owner
- Head of Customer Success
- Notes
- Browser extension that records and summarises client calls.
- Regulated context
- Tools that record or transcribe conversations can trigger recording-consent, retention and privacy rules, depending on where the participants are. Readiness mapping, not certification: this flags where to look, based on the tool's name and notes; it does not assess compliance.
ApprovalNot approved - low
Copilot
Engineering
- Account
- Business tier
- Data
- Internal
- Why this rating
- Business-tier account used with internal data only.
- Recommended action
- No action needed. Re-check at the next quarterly review.
- Owner
- Engineering lead
- Notes
- Code completion for the engineering team.
ApprovalApproved - high
AI resume screener
HR
- Account
- Not confirmed
- Data
- Confidential
- Why this rating
- Account tier isn't known and this tool touches confidential data — can't confirm what data-handling terms apply.
- Recommended action
- Confirm the account tier this week and keep confidential data out of the tool until you know.
- Owner
- Head of HR
- Notes
- Ranks applicants for open roles.
- Regulated context
- Hiring and employment-screening tools are often covered by employment, anti-discrimination or automated-decision rules, depending on where you operate. Readiness mapping, not certification: this flags where to look, based on the tool's name and notes; it does not assess compliance.
ApprovalUnder review - medium
Grammarly
Marketing
- Account
- Not confirmed
- Data
- No sensitive data
- Why this rating
- Account tier isn't confirmed — worth a quick check even though no sensitive data is known to be involved.
- Recommended action
- Confirm which account is in use (personal or company-managed), then re-rate the tool.
- Owner
- Marketing lead
- Notes
- Proofreads marketing copy.
ApprovalUnder review
| Tool | Account · Data | Risk | Approval |
|---|---|---|---|
| ChatGPTSales | Personal accountConfidential | high | Not approved |
| |||
| ClaudeEngineering | Business tierConfidential | low | Approved |
| |||
| AI note-taker (browser ext.)Customer Success | Personal accountConfidential | high | Not approved |
| |||
| CopilotEngineering | Business tierInternal | low | Approved |
| |||
| AI resume screenerHR | Not confirmedConfidential | high | Under review |
| |||
| GrammarlyMarketing | Not confirmedNo sensitive data | medium | Under review |
| |||
Report a new tool
Anything your team uses, sanctioned or not — that's the point.
Bring your own tools
Paste or upload the list of AI tools your team uses and see each one rated by the same rules as above, in the register, the policy and the reports. A plain list of names is enough to start.
What happens to your list. Reading, checking and importing it all happen in this browser tab: nothing is uploaded or saved, and it is gone when you close or reload the tab. If you then generate a policy, the tool list (names, departments, account types, data levels and approval status; not owners or notes) is sent to this demo's own server to build the draft. The public demo builds it from a fixed template and keeps nothing. Use tool names you are happy to type into a public demo.
A .csv file of up to 100 KB. It is read in your browser and shown as a preview before anything is imported. Download template
One tool per line, or a header row with columns such as Name, Department, Account type, Data sensitivity and Approval status.
What can my list look like?
Just names: one tool per line. Anything the list does not say gets the same starting answers as the form: department Unspecified, account type Not confirmed, data Internal, approval Under review.
With columns: start with a header row. The columns can be in any order and any case, separated by commas, semicolons or tabs. Put a value that contains the separator in double quotes.
- Name (required)
- Also Tool, Tool name, Application or App.
- Department
- Also Team or Dept.
- Owner
- A role or a person.
- Account type
- Personal, Business or Not sure. Also Free, Work, Company, Enterprise or Team plan. The column can be called Account or Tier.
- Data sensitivity
- Nothing sensitive, Internal, Confidential or Regulated. Also Public or Sensitive. The column can be called Data.
- Approval status
- Approved, Under review or Not approved. Also Review, Rejected or Blocked. The column can be called Status.
- Notes
- Also Comment, Purpose or Description.
Limits, the same as the form: tool name 80 characters, department 60, owner 60, notes 500; up to 100 tools on the register. A value that is too long makes that row need attention; nothing is cut. A tool name that appears twice is imported once.