Voxfra is built for agencies that deploy and manage voice AI for multiple clients. One platform, one ingestion layer, isolated tenant environments for each client.
Agencies that grow past five or six clients on custom-built infrastructure run into the same set of problems. The patterns are consistent regardless of which voice provider they use.
When agencies manage each client with a separate set of scripts and webhook handlers, every change has to be replicated manually. A prompt update, a new routing rule, or a provider SDK change becomes multiple deployments across multiple environments.
Client data sits in separate places with no consistent schema. Running a report across clients, comparing call evaluation scores, or auditing who accessed what requires manual aggregation every time.
Email notifications, CRM writes, and escalation logic are typically wired up individually for each client. When a client wants a change, it means a new build cycle rather than a configuration update.
Most agencies send clients a spreadsheet or a screenshot. There is no repeatable way to generate a structured monthly report from call data without manual work.
Each capability below is available to all clients under your Voxfra account. Configuration is shared where it makes sense and overridable per client where it does not.
Voxfra provides a single webhook endpoint per provider. Each client gets a unique slug that routes their events to an isolated tenant environment. Adding a new client is a provisioning step, not a new codebase.
Each client gets a dedicated portal scoped to their data. They can view call history, access reports, and manage their settings without seeing any other client data. The admin portal gives your team a cross-client view.
Configure evaluation criteria once and apply them across all clients or per-client. Scores are stored with each call record. Prompt evaluation lets you test configuration changes before they affect live calls.
Define rules that trigger email notifications when calls meet certain criteria. The AI extraction layer pulls relevant data from the transcript so the notification contains context, not just an alert. Rules are configured per client.
Generate an interactive report from call data for any client, covering any time period. Reports include call volume, evaluation scores, and extracted data. You can share the report directly with your client as a link.
Export up to 2,000 call records per month including transcripts, extracted entities, evaluation scores, and metadata. Useful for clients who import data into their own CRM or analytics tools.
The onboarding pattern is the same for each client you add. The goal is to make adding a new client a provisioning operation, not an engineering project.
Your Voxfra account contains your organization and the client tenants under it. Each client gets a dedicated slug and isolated database environment from day one.
Creating a client takes a few minutes. You configure the voice provider, set up the webhook endpoint, and define any email routing rules. Your client gets portal access immediately.
Set up call evaluation criteria that apply to the client. These can be shared across all clients or customized per client. Evaluation runs automatically on every call.
Each additional client follows the same provisioning pattern. Shared configuration like evaluation rubrics and email templates carries across clients. Per-client rules override the defaults where needed.
The founding plan is open to the first 20 customers. Every feature is included and the rate is locked for 12 months. We onboard hands-on and get your first client tenant live before you pay anything.