Where is our data stored?

In Quebec, on sovereign servers. Depending on the model you choose, it lives entirely on our infrastructure (SaaS), partly with you (hybrid), or exclusively in your environment (on-premises). No data leaves Quebec.

Do you use our data to train your models?

No. Data related to your organization or your callers is never used for training. We may use aggregated, anonymized statistics to improve the platform, with no way to trace them back to a person or a client.

Do you comply with Law 25?

Yes. Personal information is processed in accordance with Law 25 and the laws applicable in Canada. We apply data minimization, encryption in transit and at rest, and defined retention policies.

How does the infrastructure work?

The call arrives at our telephony gateway, is transcribed in real time and processed by our model pipeline, then the response is synthesized into speech, all in a fraction of a second.

Why open-source models rather than a big provider?

To stay in control. Starting from open models and retraining them spares us dependence on a single provider, its price increases and its policy changes. We control the agent's behaviour, the latency and where the data flows.

Is an open-source model less capable?

No, not when it is properly retrained and assembled. A closed general-purpose model is not optimized for phone conversations in Quebec French; ours is.

What happens if your service goes down?

We target 98% availability, measured monthly, with maintenance windows announced 48 hours in advance. If we fall short, service credits apply.

Does the agent understand the local accent and expressions?

Yes. It is one of the reasons for retraining: the agent is tuned to French as spoken in Quebec, not to a standardized French.

What does the agent do when it doesn't understand or doesn't know the answer?

It rephrases, then transfers to a human with the context already summarized if the request is beyond its scope. You define the escalation rules.

Can we customize what the agent says?

Completely. Script, tone, voice, qualification questions, transfer rules, information to collect: everything is configured to fit your processes.

How long does it take to get started?

In SaaS mode, a few days to a few weeks depending on complexity. On-premises hosting and custom integrations require more preparation.

Does it integrate with our CRM and tools?

Yes. The agent reads from and writes to your CRM, and connects to your tools through APIs. Existing connections are activated quickly; specific needs are developed with you.

Do we need a technical team on our side?

No. If you don't have one, we refer you to an integrator from our network who handles the connection.

How many concurrent calls can it handle?

Several parallel calls per number and several numbers per client. Capacity adjusts to your peaks.

How much does it cost?

Pricing depends on the hosting model, call volume and integrations. Contact us for an estimate tailored to your context.

Can we try it before committing?

Yes. A three-day trial lets you test the agent before any payment. Talk to our team about it.