The next named storm or burst-pipe event will collapse weeks of FNOL into days. You cannot hire for that peak. Switchboard takes the calls and texts itself and drafts the email replies, with no rota and no overtime.
The Peril Paradox
Property insurance has a flat baseline punctuated by spikes you cannot hire for. The capacity gap is structural.
A named storm or burst-pipe event can multiply your FNOL volume inside 24 hours. The handler team you sized for steady-state cannot answer those calls. Customers go to voicemail or your IVR; some hang up; some drift to social media.
Customers in genuine distress expect to be acknowledged immediately. The clock starts on cycle time, on Consumer Duty fair-treatment evidence, and on customer-experience scores. The peril does not wait for office hours.
Hiring for a once-a-year storm is uneconomic. Hiring agency staff in the middle of the event is too late. The FNOL surge has to be absorbed without a person-count change.
Even if the FNOL is captured, the surveyors, restoration contractors, and alternative-accommodation providers are also under pressure. Coordinating that supply chain on top of intake doubles the workload.
How Switchboard Absorbs It
Switchboard answers the call itself, takes the details and opens the case, so the first conversation doesn't wait for a person to come free. We don't quote a concurrency figure; the pilot is where you test it against your own peak.
Customers who can't talk during a power cut or while comforting a child can text instead. The agent has the same case tools by text as on the phone, so the details land on the case the same way.
Tell the agent which calls need a person (life-safety language, a vulnerable customer, major damage) and it puts those callers straight through to your team. Routine claims (minor escape of water, contained fire damage) go on the case for your standard workflow.
If the speech service fails during a peril event, Switchboard moves to the next provider and the call carries on. The language model is OpenAI or Anthropic, and an agent can be given the other as a fallback.
Screenshots
The sun and moon switch a screen between light and dark, and a click opens it full size.
See all 77 screens in the product tourThe Channels
Voice is the headline. Texts reach customers who can't talk, with WhatsApp on the way. Email captures the formal complaints and follow-ups.
Inbound FNOL with natural conversation and no hold music.
Text intake for customers who can't talk, with the same case tools as on the phone. Text only: pictures aren't read.
SwiftCase texts the customer as the case moves, and they can answer by replying with a code that goes straight on the case.
AI-drafted acknowledgements with handler approval before send.
Supply-Chain Calls
The bigger problem after a peril event is coordinating downstream: surveyors, contractors, alternative accommodation. Switchboard makes those outbound calls too, when SwiftCase asks it to.
Status-check calls when SwiftCase says a chase is due. The agent gets through the phone menu and writes the answer back to the case.
Confirmation and follow-up calls, one per case as SwiftCase asks for them, including properties the surveyor couldn't get into.
Supplier follow-up calls and tenancy confirmations, each answer written straight to the case.
Customers chasing for updates get a clean status conversation without consuming handler time.
FAQ
We don't quote a figure, because we haven't measured one we'd stand behind. Test it against your expected peak in the pilot, before peril season, and keep your overflow routing to people in place until you've seen it hold.
It doesn't, on its own. You write rules for it (life-safety language, repeated questions, children, medical needs, disability), and when a caller says something the rules cover, it puts them straight through to your team. It's a straight transfer, so the handler picks up a live caller with no briefing. Switchboard supports your vulnerable-customer process; it does not replace it.
Switchboard does not settle. It captures the FNOL (incident, location, third party, contact details) and creates the case in SwiftCase with the right workflow attached. Settlement decisions are human, run through your approval hierarchy with WorkflowApprovalService, and the AI never decides indemnity or coverage.
Yes, by hand. An agent's rules can be changed at any time, so you can switch to your peril rules when the event starts and back again afterwards. Nothing switches on its own: a Met Office warning won't change them for you.
Two common patterns. (1) Switchboard sits in front of the contact centre, absorbing the spike and putting through only the calls its rules say need a person. (2) Switchboard sits in the overflow, picking up calls when handler queues exceed a threshold. Either way, no platform migration is required; Switchboard just answers calls Twilio routes to it.
Yes. Commercial FNOL is structurally similar to household, with different qualifying questions, supply-chain partners, and indemnity considerations. Agent definitions are product-specific; the same Switchboard instance can run multiple agents for different product lines.
Run a 30-day pilot on overflow FNOL or out-of-hours intake. Switchboard sits alongside your existing process. When the next storm hits, you already know how it scales. No platform migration. No long lock-in.
