Help requests from the website, WhatsApp and email are read by AI, sorted by need and urgency, and opened as CRM cases for the right caseworker. Urgent cases text the on-call caseworker within 15 minutes, and every requester hears what happens next.
Typical results
15 min
Longest wait for the on-call SMS on an urgent case
100%
Requesters acknowledged with next steps
≈8 hrs/wk
Intake time on reading and sorting saved
Figures reflect typical results for this kind of automation.
The challenge
Requests for help, often long messages written under stress, arrive through a web form, WhatsApp and a shared inbox, around 300 a month. One staff member reads them in order of arrival and forwards them by hand. A family facing eviction this week can wait two days behind routine questions, and some requesters never learn their message arrived.
Before and after
Hours until an urgent case gets a call
−98%Hours per week reading and sorting requests
−80%Cases reassigned to a different caseworker
−68%The automation
AI reads each request from the website form, WhatsApp or email and sorts it by need, such as food, rent arrears, legal help or emotional support, and by urgency, such as an eviction date this week or no food at home today. It opens a case in the CRM assigned to a caseworker by area and need. Urgent cases text the on-call caseworker within 15 minutes, and every requester gets an acknowledgment with next steps.
How it works
- 01
Requests from three channels
Help requests arrive through the website form, WhatsApp and email, and each one is read the moment it lands.
- 02
AI sorts by need and urgency
AI tags the need, such as food, rent arrears, legal help or emotional support, and the urgency, such as an eviction date this week or no food at home today.
- 03
Case opened and assigned
A case opens in the CRM with a summary, the need and the urgency, assigned to a caseworker by area and need.
- 04
Urgent cases page the on-call
Urgent cases send an SMS to the on-call caseworker within 15 minutes. If nobody accepts within 30 minutes, the team lead is alerted.
- 05
Every requester hears back
Each requester gets an acknowledgment in the channel they used, saying what happens next and when to expect a call.
Starts from
- Website help form
- WhatsApp messages
- Shared email inbox
Delivers
- CRM case for caseworker
- On-call SMS alert
- Acknowledgment to requester
The impact
Urgent cases reach a caseworker in under an hour instead of around two days, and every requester knows their message arrived and when to expect a call. Intake sorting drops from about 10 hours a week to 2, and fewer cases are reassigned because they start with the right caseworker.
Questions about this automation
How does the AI decide a request is urgent?
The nonprofit defines the urgent signals with its caseworkers, such as an eviction or utility cutoff date within seven days, no food at home, or a safety concern. The AI tags these from the message text and explains why. Anything uncertain is treated as urgent, and caseworkers can change the urgency on the case at any time.
What happens to personal information in these requests?
Requests are stored only in the nonprofit's CRM, and caseworkers see the cases assigned to their area and need. The AI step classifies and summarizes the text through an API set up not to keep or train on the data. Acknowledgments contain no details of the request, only next steps.
What does the team set up before the first request is sorted?
The caseworkers agree the need categories, the urgency rules, the areas each caseworker covers and the on-call rota. On the technical side, the CRM or case system, the website form, a WhatsApp business number, the shared inbox and an SMS service are connected. This is a multi-step build, usually rolled out in stages.