Vapi alternative
The voice platform that comes with the engineers
Vapi hands you building blocks and an API. We build and run the orchestration layer ourselves, set your voice system up alongside your team, and give you a dashboard and API you own outright. EU-hosted. No lock-in.
at cost + € 0,07/min orchestration · EU-hosted · you can take the stack over
Why teams start looking
The four reasons a switch comes up
None of these mean Vapi is a bad product. They are the predictable limits of a self-serve platform once a phone line becomes something a business depends on.
The bill is hard to decompose
A single per-minute figure bundles orchestration with model and voice usage. If the model calls carry a margin you cannot see, you cannot shop them, and you cannot explain your own cost per call to a CFO.
Nobody can tell you why a call failed
When telephony breaks, a self-serve platform gives you a status field. The people who could read the SIP dialog do not work for you, and the people who work for you cannot see it.
The deepest logs are not yours
Vapi's own documentation states that system logs are strictly internal to Vapi and are never shared with customers. That is a defensible product decision. It is also the exact layer you need on the day a regulator or a customer asks what happened.
Self-hosting the alternative is a lot of work
The open-source route is real, and it comes with an on-call rota, a carrier relationship and a quarterly model review. Many teams reach for it, price the servers, and forget to price the pager.
Side by side
The differences that show up in production
We have picked the dimensions that decide how a voice system behaves when something goes wrong, rather than a feature grid where everyone ticks every box.
| What matters | A self-serve platform | VoiceDock |
|---|---|---|
| Who owns the orchestration layer | The vendor. You build on top of it through an API. | We do. We are its developers and maintainers, and we run it. |
| Who sets your system up | You do, from documentation. | We do, in working sessions with your team, then hand it over. |
| When telephony breaks | You open a support ticket and wait. | We open the SIP-trunk logs and debug at the protocol level. |
| Deep system logs | Typically internal to the vendor. | Ours to read, and ours to explain to you. |
| Where the platform runs | Usually US-hosted, with EU options varying by vendor and tier. | EU infrastructure for orchestration, dashboard, API and storage. |
| Model layer | Bundled, or bring your own key. | At cost on our keys, or bring your own, both self-serve. Local models on your own hardware are custom work we scope with you. |
| Behaviour on failure | Often a silent fallback to keep the call alive. | Fail closed, with a readable reason and a record of it. |
| Leaving | Export what the API exposes. | Take the stack over under a licence we agree with you. On-premise deployment works the same way: a conversation, not a toggle. |
| Who you are contracting with | Varies by vendor. | Flireo B.V., a Dutch company in Deventer, invoicing in euros. |
The bill
One margin, not two
Our orchestration fee is € 0,07 per minute. Model usage on our platform keys is passed through at cost, so that € 0,07 is the whole of our margin on a minute and you can see every component underneath it. Bring your own keys and the fee is unchanged while your providers bill you directly. Gemini realtime through Vertex AI is € 0,25 per minute all in. Telephony is billed per leg on top.
Where we would not argue
When Vapi is the better choice
A comparison page that claims to win on everything is worth nothing. Three cases where we would tell you to stay put.
You want to self-serve, today, alone
If your plan is to sign up at midnight, ship a prototype by morning and never speak to a vendor, a large self-serve platform is genuinely the better fit. Our funnel starts with a conversation because our delivery does.
You need the largest ecosystem and the most integrations
Scale brings breadth: more community examples, more third-party tooling, more people who have already solved your exact edge case. We are deliberately narrower and deeper.
Your volume is tiny and will stay tiny
For a handful of calls a month, engineering partnership is not what you need and you should not pay for it. Come back when the phone line starts mattering.
The wider field
The other names you should have on your list
If you are evaluating properly you are looking at more than two options. Here is the honest map, including the ones that compete with us.
Vapi
The reference self-serve builder platform. Broad, well documented, US-hosted, you assemble it yourself.
Retell AI
Positions as production-ready rather than building blocks. The strongest telephony debugging documentation of the commercial platforms.
Bland AI
Enterprise and stack-ownership angle, with the most SIP tooling on the setup side.
Synthflow
Aimed at less technical buyers, with a dashboard-first experience and its own telephony.
ElevenLabs Agents
Voice quality first. Telephony is the newer part of the offering.
Telnyx
A carrier that moved up the stack. If your problem is mostly telecom, they are a serious answer.
LiveKit Agents
Open source, and the runtime a lot of this industry is built on, including parts of ours. Self-host it and you own the operations.
Pipecat
The other major open-source agent framework. Same trade: full control, full responsibility.
We build on open foundations rather than pretending we invented the field. The difference we sell is who operates the layer and who sits next to you while it is built.
Questions
What people ask before switching
Is VoiceDock an open-source Vapi alternative?
Not as a public open-source project. What we offer instead is the practical thing people want from open source: no lock-in. The stack can be handed over and on-premise deployment exists, so you are not dependent on us continuing to exist. To be straight with you, neither is self-serve: both run under a licence and start with a conversation, because a deployment into your environment is real work rather than a setting.
What can I switch on myself after signing up?
The platform on our infrastructure: every supported provider at cost, phone and web calls, the dashboard, the API, webhooks and analysis. Local models, on-premise deployment, Trust Center access to the compliance documentation, invoicing and white-label setup are arranged with us. We publish that split on the pricing page rather than letting you find out after you sign up.
Can I migrate an existing Vapi setup?
Usually yes, and the concepts map closely: assistants, calls, numbers, tool calls and webhooks all have direct equivalents. The work is in re-pointing telephony and re-testing tool integrations, which is exactly the kind of thing we do with you rather than hand you a guide for.
What does it cost compared to Vapi?
Our orchestration is € 0,07 per minute with model usage passed through at cost. Whether that is cheaper for you depends on your provider mix and call length, so the comparison to make is total cost per connected conversation, not headline fee against headline fee.
Do you support bring-your-own-key?
Yes, and the orchestration fee is the same either way. With your own keys your providers bill you directly and you keep provider-level control over data handling.
Is my data processed in the EU?
Our orchestration, dashboard, API and storage run on EU infrastructure. The model providers you plug in are your choice, and we will tell you plainly where each one processes data rather than leaving you to find out.
What happens if a provider fails mid-call?
We fail closed: the call ends or is refused with a reason you can read in the record, rather than being silently switched to a different provider. A call that behaved in a way nobody designed is worse than a call that stopped honestly.
Bring us the call flow you have not been able to ship
Thirty minutes, no deck. We map the flow, name the provider stack, and tell you honestly whether we are the right partner for it.