Pricing

Infrastructure pricing. Not credits.

One all-in per-minute rate that includes transcription, reasoning, voice, telephony and compliance - plus a platform tier that matches how your organisation deploys. No top-ups, no coupons, no surprise provider passthroughs.

Pilot

Launch

Prove it on one queue. Production-grade from day one - same guardrails, same console.

$0.09 / min all-in
usage-based · no platform fee · cancel anytime
  • Up to 10k minutes / month
  • All pipeline stages included - STT, reasoning, TTS, telephony
  • Multi-agent calls (speak + listen + learn)
  • Knowledge grounding with citations
  • PCI redaction & consent capture
  • Email support · 99.9% SLA
Start a pilot
MOST DEPLOYED Committed use

Scale

For live queues in production. Committed minutes buy the rate down; overage at the same unit price - never punitive.

from $0.055 / min all-in
+ platform from $2k/mo · annual · invoiced
  • 100k+ minutes / month committed
  • Everything in Launch, plus:
  • Eval gates, A/B deploys & rollback
  • Campaigns with DNC wash & pre-flight
  • Live supervision - whisper & takeover
  • SSO/SCIM, scoped keys, audit export
  • 99.95% SLA · named support engineer
Model your pricing
Regulated / sovereign

Enterprise

Banks, insurers, government. Deployed in your VPC or on-prem, with your keys and your models where policy requires.

Custom · committed + platform
multi-year · security review included
  • VPC / on-prem deployment
  • BYOK models (Azure, Bedrock) & in-VPC voice
  • Region-locked residency · IRAP-aligned
  • Dedicated capacity & failover drills
  • Custom DPAs, audit rights, exit plan
  • 99.99% SLA · 24/7 with on-call bridge
Talk to sales
All-in rate - no provider passthroughs Listener & learner agents included on every call Per-queue budgets & spend caps Volume tiers auto-apply as you grow
Why teams switch

The bill you can put in a board pack.

Developer-toy platforms price like prepaid phones. Infrastructure prices like infrastructure.

What it costs youVoiceKernelCredit-based platforms
Pricing unitOne all-in $/min - STT, LLM, TTS, telephony, compliance includedPlatform fee + separate passthrough for each provider, per call
How you payMonthly invoice, PO, cost-allocation tags to your FinOps warehouseCredit card top-ups; service stops when credits hit zero
What happens at the cap Budget alert at 80% - calls never drop; overage at the same unit rate Balance exhausted mid-campaign = dead line
Multi-agent calls Compliance listener and intel agent included in the minuteEach additional model call billed separately, if supported at all
ForecastabilityCommitted rate locked for the term; per-queue forecasts in consoleRates move with each upstream provider's pricing
Cost per resolved callTracked natively - typical $0.35-0.60 vs $5-8 human blendedAssemble it yourself from logs
Questions

Pricing FAQ

What does "all-in" actually include?

Everything a production call consumes: transcription, reasoning-model inference, voice synthesis, telephony carriage (standard AU inbound/outbound), the listener and learner agents, redaction, recording, storage within your retention policy, and the console. The only extras are premium telephony routes (international, toll-free at carrier cost) and BYOK models you bill directly.

How does committed use work?

You commit to a monthly minute floor for the contract term; the rate steps down with volume. Unused minutes roll forward one month. Overage is charged at your committed unit rate - growth is never penalised.

Do you charge for failed or abandoned calls?

Connected time only, metered per second after the first 10 seconds. Unanswered outbound dials, voicemail detections under 5 seconds, and platform-fault calls are free.

What does a typical bank pilot cost?

Most pilots run one queue at 3-8k minutes for 4-6 weeks: roughly $300-700 in usage on Launch, no platform fee. You'll have containment, CSAT and cost-per-resolved-call numbers before you sign anything annual.

Is the VPC deployment priced differently?

Enterprise adds a platform fee for the dedicated control plane plus your committed usage; model inference you bring (BYOK) is billed by your provider directly. We'll model it against your call volumes in the first sales call.

Next step

Price it on your real call volumes.

Send us last quarter's call counts and handle times - we'll return a modelled bill and a cost-per-resolved-call forecast within two business days.