Dograh
Customer support

Support calls and chats tied to your helpdesk

Dograh is the self-hostable voice agent platform with an open-source workflow builder. Answer phone calls and website conversations from the tickets and help articles you already keep, with every model your team's choice.

6,000+GitHub starsBSD-2open source$0platform fee, self-hostedAny systemconnects to the tools you use
01 / How it works

Read the ticket, solve it, write it back

Example
A customer calls about a refundticket readansweredoutcome written back
01 · Look up

The ticket is open before the agent speaks

Pre-Call Data Fetchloads the record as the call connects

Your helpdesk returns open tickets for the calling number. The agent confirms identity before naming any of them.

→ A phone number finds a record, not a person

02 · Answer

Answers come from your help articles

Knowledge Baseyour own documents the agent searches

Upload policies and how-to pages as PDF, DOCX or TXT files, and the agent answers from them rather than guessing.

→ Keep each file under 5 MB

03 · Close

The outcome lands in the ticket

Webhook nodesends call results when the call ends

The summary, transcript link and answers post to your helpdesk, and failed deliveries retry on their own.

→ Use the run ID so a retry never duplicates a note

02 / Channels and systems

Where customers reach the agent

Customers reach it by
Phone on your own carrierInbound WhatsApp voice callsA voice or chat widget on your site

A website chat lasts up to an hour and shows up in call history with its transcript

Connects to
Your helpdesk's APIYour CRM or order systemn8n and similar tools

HTTP API tools, MCP tools and the Webhook node carry each read and write

Models
Dograh's own modelsYour keys for OpenAI, Google or DeepgramOpen-weight models on your servers

With a provider key, call audio still reaches that provider

03 / Why support teams pick it

A support agent your team can open up

Each step drawn on a canvas

A support lead writes each node's prompt in plain language, and the graph sets the order of the conversation.

A person when the customer asks

Your support line rings first, and the customer joins only once someone picks up.

QA on every step

Each node gets its own score, tags and summary, so you see exactly where a conversation went wrong.

No charge per resolution

Self-hosted, there is no per-minute platform fee, though servers, model keys and phone minutes cost. Dograh Cloud charges 1¢ a minute.

04 / Dograh vs support suites

What changes when the code is yours

What you getDograhhosted customer-service AI suites
Source code your engineers can read and change✓✕
Runs on servers your company controls✓✕
Your team chooses every speech and language model✓✕
No fee per conversation or per resolution✓✕
05 / Guides

Compare support agent platforms in depth

Questions
Which helpdesks does it connect to?+

Any helpdesk with an API, such as Zendesk or Salesforce. HTTP API tools read and update tickets during the conversation, and the Webhook node posts results afterwards.

Does it answer email too?+

Email stays with your helpdesk. The agent covers phone calls, inbound WhatsApp voice calls and a voice or text widget on your website.

Can the agent hand a customer to a person?+

Yes, on Twilio, Telnyx, Plivo, Cloudonix and Asterisk ARI. Write the customer's details to the ticket first, so your team picks up already informed.

Can we keep customer data in-house?+

Self-host the platform and run open-weight models on your servers, and call audio stays inside your network. Provider keys send audio to that provider.

How many conversations can run at once?+

A self-hosted install runs 10 calls at once by default, and you can raise it. Dograh Cloud includes 10, with more on request.

Can we test it before customers hear it?+

Yes. Run test calls from the editor, then call from a real phone, since most phone audio arrives at 8 kHz while a browser runs at 16 kHz.

Put your busiest support question on it first.

Self-host in one command, or start on Dograh Cloud.

$ curl … start_docker.sh && ./start_docker.sh