Policies in Incidents

When a staff member records an incident, Signal’s AI doesn’t just help classify the concern — it also identifies which of your school’s policies are relevant. This ensures staff are guided to the correct procedures without having to search for the right document themselves.

​

How AI matches policies to incidents

When an incident is submitted, Signal’s AI analyses the description and compares it against all the policy documents your school has uploaded. The AI looks at the content of both the incident and the policies to determine which ones are most relevant. For example, if a staff member records a concern about online bullying, the AI might match it to your school’s anti-bullying policy, online safety policy, and safeguarding policy.

​

What you see

When viewing an incident, you see a list of the policies that the AI has matched. For each matched policy, Signal shows:

• The policy name — so you can quickly identify which document applies
• Relevant excerpts — key passages from the policy that relate to the incident, so you can see the applicable guidance at a glance

This saves time and gives staff immediate context on what procedures to follow.

​

Viewing the full policy

If you need more detail than the excerpts provide, you can click through to view the full policy document. This opens the complete document so you can read the relevant sections in full and follow the procedures laid out by your school.

​

How matching works

Signal uses semantic embedding and vector similarity to match policies to incidents. This means policies are matched based on conceptual meaning rather than exact keywords — the system understands the intent behind both the incident and the policy. For example, an incident describing a child being left unsupervised would match a neglect policy even if the word “neglect” does not appear in the description. Key factors that influence matching:

• The incident description — more detailed descriptions lead to better, more specific policy matches
• The policy content — policies with clear, detailed text are easier for the AI to match accurately
• Policy categories — the category helps the AI narrow down which policies are most likely relevant

Matching happens automatically each time an incident is created or re-analysed. You do not need to select policies manually.

​

Benefits