Documentation

Helpdesk AI knowledge base.

Everything you need to understand, set up, and get the most out of Helpdesk AI. Use the navigation on the left to jump to any section.

How it works

Understanding the call flow

Helpdesk AI sits between your existing support phone line and your team. It answers inbound calls, collects the important context, and delivers a clean handoff so your staff always have something actionable to work from.

Call forwarding

Helpdesk AI works by routing inbound calls to a dedicated number provisioned through the platform. You can either forward your existing support line to this number or use the Helpdesk AI number directly as your published contact number.

Forwarding is configured at the carrier or PBX level and does not require any changes to your internal systems. Once forwarding is active, every inbound call arrives at your Helpdesk AI agent rather than ringing out or hitting voicemail.

Tip: Many hosted phone systems and VoIP providers allow conditional forwarding, so you can forward only when your team is unavailable or all lines are busy. Helpdesk AI works well as a fallback layer in this configuration.

Inbound calls and your AI voice agent

When a forwarded call arrives, your custom AI voice agent answers immediately. The agent is configured specifically for your organisation and is not shared with any other customer.

During the call, the agent:

  • Greets the caller in a tone and style suited to your support operation
  • Guides a structured conversation to capture the caller's identity, the nature of the issue, and any relevant urgency signals
  • Handles clarification naturally, asking follow-up questions where the caller's initial response is vague or incomplete
  • Closes the call clearly and confirms the next step for the caller

The captured information is then structured and passed into your workflow, ready for triage, ticket creation, or escalation depending on your setup.

Setting up

Getting started with Helpdesk AI

Setup is designed to be straightforward. Most customers are live within a single working day from account creation.

Selecting a telephone number

After your account is created, you will be prompted to select a telephone number for your Helpdesk AI agent. Numbers are available in a range of regions and area codes. You can:

  • Choose a new number from the available pool directly in the dashboard
  • Request a specific area code or geographic number if needed
  • Use your existing published number by forwarding it to the number assigned to your agent

The number assigned to your agent is dedicated to your account. It will not be reassigned or shared.

Forwarding calls to your agent

Once you have your Helpdesk AI number, you need to configure your existing support line to forward inbound calls to it. How you do this depends on your phone provider:

  • VoIP / hosted PBX: Log in to your phone system admin panel and set up unconditional or conditional call forwarding to your Helpdesk AI number
  • Traditional landline: Contact your carrier and ask them to configure call diversion or call forwarding on your support number
  • Mobile: Use the network's standard forwarding codes (for example, **21*[number]# on most GSM networks) or configure forwarding through your mobile admin portal
Note: If you are unsure how to configure forwarding, contact our support team and we can walk you through the steps for your specific provider.

Billing

Payments and subscription

All billing for Helpdesk AI is handled securely through Stripe. We do not store your payment details directly.

Payment options

You can choose between monthly and annual billing at any point:

  • Monthly billing: Your subscription renews on the same calendar date each month. You can cancel or change your plan at any time from the billing section of your dashboard.
  • Annual billing: Pay for twelve months upfront and receive a discounted rate compared to the monthly equivalent. Annual subscriptions renew automatically unless cancelled before the renewal date.

All payments are processed by Stripe. Invoices are issued automatically and can be downloaded at any time from your billing dashboard. Accepted payment methods include all major credit and debit cards, and SEPA Direct Debit where available.

Enterprise billing: Enterprise customers may be eligible for invoiced payment terms. Contact us to discuss options.

FAQ

Frequently asked questions

What AI models do you use?

The models used by your agent are selected based on your workflow requirements rather than a single fixed stack. Different tasks within a call — speech recognition, language understanding, response generation, and voice synthesis — each have different accuracy, latency, and privacy trade-offs, and we match the model to the task.

We work extensively with open-source models for speech-to-text (STT) and text-to-speech (TTS), including models such as Whisper, Coqui TTS, Piper, and Kokoro. These are not only production-grade but form the basis of our internal training and evaluation pipeline. We use real-world call data (anonymised and processed in accordance with our data policy) to continuously improve the performance of our agent configurations.

Where proprietary API-based models are used for specific capabilities, this is documented in your account configuration and can be reviewed or changed. Enterprise customers have the option to restrict the agent entirely to locally-hosted, offline models — see the Enterprise section of the pricing page for details.

How can I know my data will be safe?

Data privacy is a core design requirement, not an afterthought. Here is what that means in practice:

  • EU-hosted infrastructure: All production servers and call data storage are hosted within the European Union, in data centres that comply with applicable data protection regulations including GDPR.
  • Model selection for privacy: AI models are evaluated not only for performance but for how they handle data. We prefer models that process data locally or in isolated inference environments over models that route data through shared external APIs.
  • Strict customer isolation: Your AI agent belongs to your account only. It does not share a context, training data, memory, or outputs with agents used by any other customer. There is no crossover between accounts at any layer of the system.
  • Call data handling: Call recordings and transcripts are stored only as long as required for your configuration (for example, to produce a structured handoff). Retention periods are configurable and can be set to zero if your organisation requires it.
  • Enterprise offline models: For organisations with the highest data sensitivity requirements, our Enterprise tier supports fully offline and self-hosted model deployments. In this configuration, no call content leaves your environment — AI processing happens on infrastructure you control, with no external API connections.

If you have specific compliance requirements or need documentation for an internal review, contact us and we will provide the relevant information.

I don't want to install an untrusted endpoint agent

This is a reasonable concern and one we take seriously. Helpdesk AI does not require you to install any endpoint agent on your own systems. The core product operates entirely at the telephony layer — calls are forwarded, handled, and structured without touching your endpoints.

For customers who use a remote monitoring and management (RMM) platform and want Helpdesk AI to push structured call data directly into their RMM workflows, we support a range of integration methods that do not require an agent installed by us:

  • Webhook delivery: Helpdesk AI can POST structured call data to any webhook endpoint your RMM or ticketing system exposes. You configure the endpoint; we send the data to it.
  • Your own RMM integration: If your RMM has an API, you can use it to pull data from Helpdesk AI on your own schedule, using credentials and access rules you define.
  • Email handoff: As a simple baseline, structured call summaries can be delivered by email to any address or distribution list your team controls.

Where an agent-based integration is used as an option, the command palette available to the Helpdesk AI agent is strictly limited. It can only perform the specific actions you have explicitly enabled in your account configuration — it cannot browse the filesystem, execute arbitrary commands, or take actions outside the defined scope. The full list of permitted actions is visible and auditable in your dashboard at any time. For more on how Resolve handles endpoint interactions, see the Resolve product page.

Still have concerns? Our team is happy to walk through any specific integration requirements, answer technical questions about how data flows, or arrange a call with an engineer if needed.