Your accounts, your keys
Runs in your company's cloud, with hosting, telephony and model-provider accounts in your name. You hold admin access and set permissions, retention and exports.
A brief for PE owners and portfolio CEOs
At 30,000 call minutes a month, versus outsourced agents. We build a production voice agent in your company's own cloud, telephony and model accounts and run it 24/7 for a flat fee. Providers bill at cost, with no markup on minutes.
Projections, not reported HorsePower results, for an equivalent, AI-suitable calling workload over 12 or 36 full operating months. First-year and three-year figures deduct the setup fee once. Excludes taxes, migration overlap and incremental human exception handling.
01 — The economics at 30,000 minutes a month
| Saving compared with | Per month | First-year net | Three-year net | Setup payback |
|---|---|---|---|---|
| Outsourced live agents | $14,500 · 60.4% | $144,000 | $492,000 | 2.1 months |
| Hosted voice AI benchmark | $5,500 · 36.7% | $36,000 | $168,000 | 5.5 months |
Projections, not reported HorsePower results, for an equivalent, AI-suitable calling workload over 12 or 36 full operating months. First-year and three-year figures deduct the setup fee once. Excludes taxes, migration overlap and incremental human exception handling. Benchmark basis in the fine print below.
02 — What your company owns
Most voice AI is bought as a service. We deliver a capability your business controls, then manage it for you.
Runs in your company's cloud, with hosting, telephony and model-provider accounts in your name. You hold admin access and set permissions, retention and exports.
Scripts, qualification rules, escalation logic and integrations are assigned to your company, with source code, configuration and documentation delivered.
Every call record is available for your own quality reviews and analytics. You approve which speech, model and telephony providers may process calls.
A handover package, access register and tested exports mean a new operator, or a new owner, starts from documented rules rather than rediscovery.
03 — Why it matters to an owner
Providers bill infrastructure at cost and our fee is separate, so the board sees exactly what a minute of calling costs.
Service rules live in documented code your company holds, not in an outsourcer's training binder or a vendor's dashboard.
Approved components can be reused where systems match. Each company keeps its own data; new brands are scoped separately.
04 — Proof
French, English and Italian support with ERP order lookups, warranty claims opened on the call and qualified leads written into Zendesk. Launched on one call in five, then scaled.
VoxDonna built this managed voice AI system for HorsePower Brands: customer-owned infrastructure, open orchestration and managed day-to-day operation.
The savings in section 01 are projections for an equivalent workload, not reported HorsePower results.
Le Marquier figures are VoxDonna-published (voxdonna.com/case-studies/le-marquier), not independently audited. Control means governing where data is stored, who can access it and which providers may process it; those providers still process call data under their own terms. Third-party software and model IP stay under their licenses. No resale value or valuation uplift is assigned to the asset. Hosted platforms can also support customer-owned prompts and data; our distinction is customer control of the deployment plus managed operation and handover.
05 — What we build, what we run
06 — Pricing as volume grows
| Monthly minutes | Infrastructure, at cost, paid direct | VoxDonna management | Monthly total | Effective rate | One-time setup | Monthly saving vs $0.80 outsourced |
|---|---|---|---|---|---|---|
| 10,000 | $2,500 | $2,000 | $4,500 | $0.45 / min | $10,000 | $3,500 |
| 30,000 | $5,500 | $4,000 | $9,500 | $0.317 / min | $30,000 | $14,500 |
| 50,000 | $8,500 | $6,000 | $14,500 | $0.29 / min | $50,000 | $25,500 |
Indicative tiers; final price depends on scope. New brands, locations, workflows, integrations and custom analytics are quoted separately.
07 — How we prove it before you switch
Review your current invoice, eligible call types, outcomes and escalation rules.
Run a defined subset of calls alongside your current setup.
Agreed thresholds for completed requests, quality, latency and reliability.
Move volume step by step with rollback ready; confirm access, exports and handover before sign-off.
The business case includes human exception handling and any costs retained with the incumbent. We do not assume every call can be automated.
Next step · 30 minutes
In 30 minutes we confirm your real cost per minute and pick one workflow to pilot. Then we prepare a scoped pilot proposal built on your numbers, not ours.
$0.80/min is VoxDonna's indicative rate for US outsourced live agents; our published cost model (voxdonna.com/blog, Sept 2026) puts US agent labor alone near $0.64 per handle minute, with all-in contact-center costs roughly double. The $0.50/min hosted-AI benchmark is all-in pricing reported to us by US clients and consultants, not any platform's published rate. VoxDonna figures follow its indicative managed-service proposal; amounts in USD unless marked. Savings = minutes × comparator rate, less VoxDonna infrastructure and management. Confirm equal service scope and quality before treating any difference as realizable.