An IVR people can talk to, and a bot that knows when to stop.
Every inbound number becomes a campaign with its own IVR, queue behaviour, routing rules, staffing plan and SLA. Build the flow visually and publish it without a maintenance window.
Most IVR is bad because it is built once and never revisited
The complaint about phone menus is rarely about menus. It is about menus reflecting an org chart from four years ago, offering nine options when callers want two.
The fix is not removing the IVR. It is making the flow cheap enough to change that it tracks reality. Firetone's is a canvas, published without taking the number offline.
Building and routing the call
Visual IVR designer
Menus, prompts, data dips and conditional branches assembled on a canvas, published without a maintenance window.
Skills-based routing
Match on language, product, account value and agent proficiency rather than dealing calls round-robin.
Value-based routing
Route by what the account is worth or what the call is about, so the records that need a senior agent reach one.
Queue callback
Offer a call back instead of hold time while keeping the caller's position in the queue.
Screen pop
The customer record is open before the agent says hello, with the IVR path and any bot conversation attached.
Blended agents
Move seats between inbound and outbound as volume shifts, without moving them between systems.
Live wallboards
Service level, longest wait and abandon rate per campaign, visible while the shift is running.
Campaign SLA reporting
Each inbound number carries its own staffing plan and SLA, reported against separately.
How a conversational call resolves
The caller says what they want
Instead of mapping their problem onto a numeric menu, the caller states it in natural speech. The bot captures intent rather than a keypress.
The bot attempts the routine case
Balance enquiries, booking changes, delivery status and similar requests are completed end to end, including the data lookups they require.
It decides whether it can finish
Where intent is unclear, the request is out of scope, or the caller asks for a person, the call escalates.
A person picks up with context
The agent receives the transcript of what the bot already established, the customer record on screen.
Deflection is the wrong thing to optimise
Deflection — calls resolved without a person — is easy to measure and easy to improve dishonestly. A bot that makes escalation hard posts excellent numbers and a worsening relationship.
The better question is whether the deflected calls genuinely did not need a person, and whether escalated ones arrived better prepared. That is why the handoff threshold is yours to set.
Included with inbound campaigns
- Visual IVR designer with menus, prompts, data dips and conditions
- Publish flow changes without a maintenance window
- Conversational voice bot with natural-speech intent capture
- Configurable bot-to-agent handoff thresholds
- Skills, priority and value-based routing
- Callback in queue with position retained
Every inbound call is scored, not sampled
Bot and agent calls alike are transcribed and scored against your rubric — all of them, not the sample a QA team can listen to. Live sentiment escalates a deteriorating call while it is still open.
IVR and voice bot questions
What is the difference between a traditional IVR and a conversational one?
A traditional IVR presents a menu and collects keypresses, so the caller has to map their problem onto the options you thought of. A conversational IVR captures intent from natural speech, which means callers can state an unusual request directly instead of pressing zero until they reach somebody.
Can we change the IVR flow ourselves?
Yes, and this is the point of building it on a canvas. Menus, prompts, data dips and conditional branches are edited visually and published without taking the number out of service.
How does skills-based routing differ from round-robin?
Round-robin deals the next call to the next available agent regardless of fit. Skills-based routing matches on attributes — language, product knowledge, account value, agent proficiency — so the call reaches somebody equipped to resolve it.
What happens when the bot cannot help?
It escalates to an agent, and the handoff carries context: the transcript of the conversation so far, the intent it captured, and the customer record. Routing uses that captured intent, so the caller does not repeat themselves.
See it running on your own traffic.
The free trial includes one module, ten seats and the full AI feature set, with an engineer configuring your first campaign alongside you.