At a glance
| Client | National Emergencies Trust |
|---|---|
| Sector | Charity and non-profit, emergency response |
| Service | Agentic AI Workflow Design |
| Evidence base | Six-pillar framework and a library of more than 130 historical emergencies |
| Result | Board-ready assessment in under 15 minutes, down from several hours |
| Use since 2025 | About a dozen major incidents monitored and triaged |
Why did the charity need faster emergency assessments?
When a major incident happens, trustees have to choose fast. They can launch a national appeal, support the local response, or keep monitoring. Each choice carries cost and reputational risk.
Before this engagement, staff gathered the evidence by hand under pressure. It took hours. The method also varied from one emergency to another, so trustees could not compare like with like.
The charity needed evidence that was fast, consistent and traceable. It did not want a tool that made the decision for them.
What did Insightful AI build?
We designed and built a governed decision-support agent. An agent is an AI system that follows a set process, gathers information and produces a defined output. This one assesses an incident against six pillars of evidence:
- Impact: scale of harm to people and places
- Urgency: how fast the situation is changing
- Resource strain: pressure on local responders
- Public attention: media and public interest
- Aid gaps: where existing funding and support fall short
- Mobilisation potential: whether an appeal would succeed
The agent draws on two sources. The first is a structured library of more than 130 documented historical emergencies. The second is live official sources for the incident in progress. It compares the new incident with past ones and ends with a recommendation.
How does the governance work?
The agent informs decisions. It does not make them. Five controls apply to every assessment:
- Confirmed and unconfirmed information are separated in every output
- Every source is logged with a timestamp
- The recommendation states its evidence and its limits
- Trustees make the call and record it
- The six-pillar method is fixed, so assessments stay consistent across incidents
This maps to the third stage of our Adoption Roadmap: establish governance. It follows the People, Process, Principles model that shapes every build.
What were the results?
| Measure | Before | After |
|---|---|---|
| Time to a board-ready assessment | Several hours | Under 15 minutes, as reported by the team |
| Incidents supported since 2025 | Manual, one at a time | About a dozen major incidents |
| Method | Varied by emergency | Fixed six-pillar framework |
| Source trail | Gathered by hand | Every source logged with a timestamp |
In one recent case, a full assessment was ready within 100 minutes of a major incident.
What we learned
Consistency matters as much as speed. Trustees valued a fixed method because it let them compare incidents. A faster but inconsistent tool would have solved half the problem.
Historical cases anchor the judgement. A library of past emergencies gives the agent something to compare against. Without it, every assessment starts from zero.
Keep the decision human. The agent recommends. People decide and stay accountable. This is what made trustees comfortable using it.
Who is this for?
This approach suits any organisation that makes high-stakes decisions under time pressure:
- Grant-makers and emergency appeal funds
- Charities that triage requests for help
- Public sector teams with incident response duties
- Boards that need evidence packs at short notice