School Trip Risk Assessment: A Template Drafted With AI, Signed by You

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for School Trip Risk Assessment: A Template Drafted With AI, Signed by You.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for School Trip Risk Assessment: A Template Drafted With AI, Signed by You.

Write the trip brief, ask AI to draft the hazards, who is at risk, existing controls and residual risk in a table, then walk the route, read every line against the real venue and your pupils, change what is wrong, and sign it yourself. AI speeds the first draft; it never replaces your judgement or your signature.

A risk assessment is a record that a named person thought about what could go wrong and decided what to do about it. If AI writes every line and nobody tests it, you hold a polished document and no assessment. The template below is built to keep the thinking with the leader.

Follow me on Instagram@sagnikteaches

What a trip risk assessment actually contains

Whatever format your school uses, the core is the same five columns. Learn them and the document stops being a form and becomes a way of walking through the day.

Connect on LinkedInSagnik Bhattacharya
  • Hazard. Something with the potential to cause harm: a road crossing, a wet floor, an animal, a lost child.
  • Who might be harmed. Pupils, adults, the public, and which pupils in particular (those with mobility needs, medical plans, a history of absconding).
  • Existing controls. What you already do: one adult per ten pupils, seatbelts, hi-vis tabards, a head count at every transition.
  • Risk rating. Likelihood times severity, before and after controls (residual risk). Most schools use a simple low, medium, high scale or a five-by-five grid.
  • Further action, owner and date. What is still to be done, by whom, before the trip.

A school will usually also record the venue and date, group size, staff names, the emergency plan, a weather decision point, and the review and sign-off. Your own visits policy decides the exact layout, and where it sets one, use it and let AI work inside it.

Subscribe on YouTube@sagnikteaches

Start from a trip brief, not a blank prompt

The quality of an AI draft depends on how specific the input is. "Write a risk assessment for a school trip" returns a generic list that could be about anywhere. A brief that names the venue, the travel and the group returns hazards you can recognise. The same brief can also feed the parent pack, as set out in the school trip letter, consent slip and kit list tutorial, so the two documents cannot disagree.

A filled-in brief for the worked example (all details illustrative):

TRIP BRIEF FOR RISK ASSESSMENT
Group: Year 5, 54 pupils, ages 9-10. Two pupils with allergy plans (adrenaline pens); one pupil uses a wheelchair; one pupil with a history of leaving the group.
Destination: a science museum in a city centre, 38 miles away.
Travel: one 70-seat coach each way, operator the school has used before; drop-off and pick-up at the museum's coach bay on a one-way street.
Walking: none, except from coach bay to museum entrance (about 80 metres, one pedestrian crossing controlled by lights).
Day: depart 8:30, arrive 9:45, workshop 10:30, lunch 12:15 in the museum's school room, free-flow galleries in groups 13:00-14:30, depart 14:45.
Staff: 6 adults (4 school staff, 2 parent helpers), ratio 1:9 (six groups of nine). Trip leader holds first-aid certificate.
Venue: busy; public galleries; lifts and stairs; shop at exit.
Weather: indoors, but queueing outside possible.
Communication: leader's mobile; each group leader has a mobile; school office contact for emergencies.
Emergency meeting point: museum main entrance foyer, beside the information desk.
The Complete Excel Course with AI Integration thumbnail

The Complete Excel Course with AI Integration

From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.

By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner

Buy now

The first-draft prompt

Ask for a table, in the exact columns your school uses. State what the AI must not do, because the failure mode is confident invention.

Draft a school trip risk assessment table from the brief below.
Columns: Hazard | Who might be harmed | Existing controls | Likelihood (1-5) | Severity (1-5) | Residual risk (low/medium/high) | Further action (owner, by when).
Cover: travel by coach (loading, seatbelts, travel sickness), pedestrian crossing, head counts and a missing pupil, crowding in galleries, stairs and lifts, allergies and medication, pupil with mobility needs, free-flow gallery periods, lunch and choking/allergen risk, toilets, weather while queueing, lost or separated groups, emergency contact and first aid, behaviour and supervision, medical emergency, and delays or coach breakdown.
Rules: use only the facts in the brief. Where information is missing, write "LEADER TO CONFIRM" in the Existing controls column. Do not state any venue facility, price or rule not in the brief. Do not claim any control is in place unless the brief says so. Give 14 to 18 rows.

[paste brief]

The line "LEADER TO CONFIRM" is the important one. It turns the AI's guesses into a to-do list for you.

What the first draft looks like, and where it is wrong

Three rows from a realistic first draft (illustrative):

HazardWhoExisting controlsLSResidual
Pupils crossing from coach bay to entrancePupils, adults, publicCross at the lit crossing only; adults at front, back and kerb side; hi-vis tabards on adults25Medium
Pupil separated from group in galleriesIndividual pupilsGroups of 9 with a named adult; head count at every room change; pupils carry a card with the school's emergency number33Medium
Allergic reactionTwo pupils with plansLEADER TO CONFIRM: medication carried by named adult25Medium

Where a draft like this usually goes wrong:

  • Invented controls. "Pupils wear high-visibility vests" appears, but the school owns eight vests for 54 pupils. Say what you actually have: adults only, or you borrow more.
  • The hazard is real, the control is generic. "Adults will supervise" is not a control. Replace with who, where, when: "Mr [name] stands at the crossing control; Ms [name] at the rear; head count on the pavement before crossing."
  • Severity scores glossed. Crossing a road scores 5 for severity even with 2 for likelihood. A product of 10 is high on the grid used below, not the "Medium" the draft gave it, and the allergy row has the same slip. Be honest, and add actions that bring it down.
  • Missing the real worry. The row for the pupil with a history of leaving the group was written for "individual pupils" generally. Make it specific to that pupil's plan, with a named one-to-one adult and a briefing. Use initials in the working copy, not full names, until you move out of the AI chat.

The worked coach-trip example, row by row

Let's finish a few rows properly, as a leader would, with a score and an action. Ratings use a five-by-five grid where residual risk of 1 to 4 is low, 5 to 9 medium, 10 or more high (illustrative; use your school's own scale).

  1. Coach loading on a one-way street. Hazard: pupils step off the kerb into traffic while boarding. Existing controls: coach stops in the museum's marked bay, adults hold the line on the kerb side, pupils board and alight one at a time on the pavement side only. Before controls: L3 x S5 = 15. After: L1 x S5 = 5, medium. Action: leader phones the coach operator by 5 March to confirm the stop position and the driver's name.
  2. Pupil in a wheelchair, stairs and lifts. Hazard: separated from group if a lift is out of order. Controls: venue's access information checked by phone; step-free route agreed; a named adult stays with the pupil; groups plan their route to use the same lift. L2 x S3 = 6 before; L1 x S3 = 3 after. Action: leader asks the museum for its accessible route map and shares it with the group's adult.
  3. Allergy emergency. Two pupils with plans. Controls: medication and a copy of each plan carried by the named first-aider, not left in a coach hold or a bag in the lunch room; staff briefed on the signs and on calling an ambulance; lunch eaten in the school room. L2 x S5 = 10 before; L1 x S5 = 5 after. Action: medical lead briefs staff on 10 March; leader checks that the in-date pen is physically in the first-aider's bag on the morning.
  4. Missing pupil. Controls: head counts at coach, entrance, before and after lunch, before each gallery change, and before the coach leaves; group sizes of nine; named adult per group; a plan agreed in advance of who stays with the group and who goes to the information desk. Action: the leader gives each adult a printed list of their group and the leader's mobile number.
  5. Delay or breakdown. Controls: coach operator's breakdown number on the leader's sheet; the school office will inform parents using the existing text system if the return is more than 30 minutes late; a pre-written message is ready. This ties into written weather and welfare decisions made before the day, so nobody drafts a parent message in a panic.

Notice that every action has a name and a date. An unowned action is a wish. When you ask AI to tidy your assessment, instruct it never to remove an owner or a date.

The leader walk-through

You do not need to be a safety professional to catch the AI's mistakes. You need to be physically or virtually at the place. Do the following before you sign.

  1. Visit or video-call the venue. Many schools require a pre-visit; where it is not possible, speak to the venue's education team and ask for their own risk assessment. Walk the coach bay to the entrance, the toilets, the lunch room and the exit route.
  2. Read the table out loud with a colleague. A second adult asks "who exactly does that?" for each control. If nobody can say, rewrite it.
  3. Check each control is true today. Hi-vis vests: counted. First-aid bag: packed. Mobile numbers: dialled. Lift access: confirmed.
  4. Cross-check against the pupil list. For every pupil with a health plan, behaviour plan or access need, confirm there is a named adult and a line in the assessment. This is done using the school's own records, outside the AI tool.
  5. Brief the staff. Give every adult a one-page version of their role: group list, route, head-count points, emergency contacts and what to do if a pupil is missing.

The same habit of a human reading an AI draft against reality is behind the checks in the client report checklist for AI-written reports: the document is only as good as the checks done by the person who signs it.

Why AI cannot sign it off

There are three practical reasons beyond the principle. First, an assistant has not seen the venue. It combines patterns from similar documents and presents them in the register of someone who has. Second, it does not know your pupils, and the highest-consequence risks on a trip are usually specific to a child. Third, accountability sits with a person. If something goes wrong, the question asked will be who assessed the trip, what they did and whether the controls were in place. "The tool said it was fine" is not an answer.

A safe rule to write in your school's policy: AI may produce a first draft or a formatted copy; the trip leader must review and edit every row; the approver signs a document that names a human author. Keep a line on the form, "Drafted with AI assistance, reviewed in full by [name] on [date]". It is honest, and it reminds the signer what the signature means.

Data to keep out of the tool

The brief above includes a pupil with a medical plan and a pupil with a history of leaving the group. Even without names, that can identify a child in a small school. Work like this:

  • Describe needs in general terms in the AI chat ("two pupils with allergy action plans; one wheelchair user; one pupil needing close supervision").
  • Keep names, photographs, health details and plan contents in your own system.
  • Add the pupil-specific lines after you have left the chat, using initials or your own codes.
  • Use an account your school has approved. OpenAI, Anthropic and Google each say their business plans do not use your content for training by default, but check the vendor's current terms for the plan you use.

Different trips, different drafts

The method stays; the hazards change. Three quick variations (illustrative):

  • A walk to the local library. Two or three rows: road crossing, group control on pavements, public library behaviour. The AI draft should be short. If it returns 16 rows for a walk, it is padding, and you should cut.
  • A residential trip. Add rows for overnight supervision, bedrooms and fire procedures, activity providers, medication storage, homesickness and safeguarding arrangements. Ask the provider for their own documents and read them; a provider's paperwork does not replace yours.
  • A sports fixture at another school. Add changing arrangements, minibus or parent-car transport rules, and pitch conditions. If you use parent drivers, your visits policy will have specific checks, and these should not be improvised from an AI draft.

Event teams handle a similar problem of timed groups, roles and contingencies; how event planners use AI to build budgets and run sheets is a good model for the day plan that sits beside the assessment.

Scoring honestly: a quick sum on one row

Scores cause more arguments than hazards, so agree the scale before you start. Take the lift row. Likelihood: the museum has two lifts and one has been out of order twice this year, so a leader might say 3. Severity: a pupil is delayed or separated from the group, nothing worse, so 2. Before controls the product is 6, medium. With a named adult staying with the pupil, a phone number for the venue's access team and a plan to meet at the foyer if the lift fails, likelihood of harm drops to 1, so the product is 2, low. Write the reasoning in the row, one short sentence, because the next leader will ask why you chose 3 and the answer should be on the page.

Using AI as a challenger, not an author

Once you have written your own version, a second prompt is often more valuable than the first. Paste your finished table and ask the assistant to find what is missing. It will not know your venue, but it is good at running down a checklist of ordinary hazards you may have overlooked on a tired Thursday.

Below is a school trip risk assessment I wrote. Do not rewrite it. List up to ten hazards or situations that a careful reviewer might ask about and that I have not covered, as questions. For each, say which row it relates to, if any. Do not state facts about the venue.

[paste finished table]

A realistic reply (illustrative) might ask: "What happens if the weather turns during the queue outside? Is there a plan for a pupil who needs the toilet during a head count? What is the arrangement if one adult has to leave the group with an ill pupil?" Each question is one for you to answer from knowledge of your trip. The third is a real gap in many assessments: if the first-aider leaves with a pupil, a group of nine is suddenly with fewer adults than the ratio. A good answer is a named reserve adult and a rule that two groups merge while the other adult is away.

A mistake that looked fine on paper

Say a draft control reads "Pupils will wear name badges with the school's phone number". It reads well and ticks the lost-child box. On the pre-visit, the leader notices the museum asks schools not to display pupils' names. The control quietly becomes a data problem, and the fix is a card with the school number and a group colour, carried inside a pocket, with no name at all. Nobody could have caught that from a table. They caught it because someone went and asked. That small story is the reason the visit stays in the process, however good the draft looks.

Review on the day and after

A risk assessment is live. On the morning, the leader checks whether anything has changed: weather, a pupil absent or newly medicated, a staff member off sick, a road closure. If so, amend by hand, initial and date it. After the trip, a five-minute debrief records what the assessment missed. Perhaps the queue outside was longer than expected, or the lift was out, or a group got ahead of the adult. Add it to the template for next time.

A short prompt can help you tidy the debrief into an improved template, as long as it is working from notes you wrote:

Below are my debrief notes from a school trip and the original risk assessment table. Suggest up to five rows to add or change for next time, and say which note each change responds to. Do not remove any existing row.

Then you decide which suggestions go in. Over two or three terms, the template improves from real experience, which is the single best advantage it can have over a generic one.

How long the coach-trip assessment takes

For the coach-trip example, expect about 20 minutes to write the brief, 10 minutes for the first draft, an hour to correct it against the venue information, 30 minutes for the walk-through and colleague read, and 20 minutes to brief staff. That is just over two hours, compared with half a day starting from a blank page. A free or low-cost plan on any mainstream assistant is enough for this task, provided no named pupil data goes in. If you want help setting up templates and approval steps across all your visits, that is the sort of thing my AI implementation consultation covers; the template above works on its own.

Before you sign: a final checklist

  1. The brief names the venue, travel, group, staff, timings and emergency meeting point.
  2. Every row has a hazard, who is at risk, a real control, a score and a residual rating.
  3. Every "LEADER TO CONFIRM" has been resolved by a call, a visit or a document.
  4. Every control is something that will physically happen, with a named person.
  5. Pupil-specific needs are covered, from the school's own records.
  6. Residual high risks are either reduced further or the trip plan changes.
  7. The emergency plan includes contacts, a meeting point and a message to parents already written.
  8. The approver has read it and signed; the form states that a human author reviewed an AI draft.
  9. Every adult has a one-page briefing and the leader's number.
  10. Morning-of changes are noted, and the debrief updates the template.

Questions about AI-drafted trip risk assessments

Can AI write a school trip risk assessment for me?

It can draft the structure and a first list of hazards and controls, which saves time. It cannot know your venue, your pupils or your staff, so it cannot judge them. The trip leader must check each line against reality, edit it, and sign. An assessment nobody has read against the site is a document, not a safeguard.

How detailed should a school trip risk assessment be?

Proportionate to the trip. A local walk may need a short generic form plus a route note; a coach trip to a farm or a residential needs a specific assessment with named hazards, controls, owners and an emergency plan. If your visits policy sets a format, use it, and only use AI to fill it in.

Who should approve the assessment?

Follow your own visits policy. In most schools the leader writes it, an educational visits coordinator reviews it, and the head teacher or equivalent approves it before parents are told the trip is confirmed. AI is never an approver or a second reviewer.

Should we use the venue's own risk assessment?

Ask for it and read it, but do not rely on it alone. It covers the venue's hazards, not your group's: your pupils' needs, your travel, your staffing and your weather plan. Use the venue's document as an input to yours, and do a pre-visit, in person or by video, where the policy asks for one.

Further reads

Sources: general educational-visit practice (hazard, who is at risk, controls, residual risk); the school's own visits policy and employer guidance; OpenAI, Anthropic and Google business privacy pages (checked 4 November 2026).

Want safety paperwork templated without cutting corners?

On a 1:1 call we can look at the safety and compliance documents your team rewrites each term, set up templates that AI can fill in as a first draft, and agree which checks and signatures stay with named people.

Book a 1:1 call with me