This Open-Source Tool Replaces Vapi for Voice AI (Dograh)

EnglishالعربيةDeutschEspañolFrançaisहिन्दीBahasa Indonesia日本語한국어PortuguêsРусский中文
BBetter Stack
Computing/SoftwareSmall Business/StartupsInternet Technology

Transcript

00:00:00you just built a voice ai agent it works then the bill shows up and you're paying for the llm the
00:00:05voice the phone call and then another platform fee on top of that that's not even the worst part
00:00:10the worst part you still don't really even own the system today i'll show you dogra
00:00:16and an open source vappy alternative you can self-host inspect and control
00:00:26voice ai nowadays can look somewhat simple from the outside take a phone call turn speech into text
00:00:33send it to the llm turn the answer back into speech it's done that's easy right well as any of us know
00:00:39who've tried this not really because real calls are messy people interrupt people go silent they're
00:00:46going to change topics they can ask really weird questions your agent needs to call apis and when
00:00:53it breaks you need to know why that is where most voice ai projects become more of a pain a voice
00:00:59agent is not just chat gpt with a phone number it is a live system with a bunch of moving parts right
00:01:06that's speech to text llm text to speech state tool calls a boatload of other things you get it there's
00:01:12a lot of moving parts that we don't actually see happening and when the call fails the bot gave a
00:01:17bad answer is not enough was it the prompt was it the model what was it why did it fail and this is
00:01:23where dogra comes in if you enjoy coding tools that speed up your workflow be sure to subscribe we have
00:01:29videos coming out all the time all right now let's look at this in practice i'm going to start locally
00:01:34because if a tool says it's built for devs i want to see docker before anything else this was super easy
00:01:39to spin up i'm going to clone it from github i'm going to cd into the folder and then i just have
00:01:44to run docker compose up that's simple enough easy enough for us once the containers are running we
00:01:50can jump into the dogra ui now i'll build a simple lead qualification agent so what do i mean by that
00:01:57someone's going to call in the agent asks what they want to build then it asks about the company the
00:02:03size the budget small things like this it'll then call an api tool to create or update a crm lead if
00:02:11we embed that and maybe i could even say if the lead was qualified it transfers to a human so i add a
00:02:18prompt node then a qualification step then an api tool call and then i can add a branch and a transfer
00:02:28there's no custom orchestration code yet and that's kind of the point here
00:02:32this looks like a no code canvas but for devs and the value is not no code the value is not wasting
00:02:39code trying to tie everything together now let's try to run a test call here hi this is sarah from
00:02:46inbound calls are you still there we're looking for an ai phone agent for inbound demo requests that's
00:02:55great i can definitely help you with that to make sure i connect you with the right solution could
00:03:00you tell me a little more about what you're looking to achieve with an ai phone agent for your inbound
00:03:05demo requests let's say around 20 000 minutes thanks for sharing that and what is your company
00:03:11size and industry now we can see the transcript here we can see the trace we can see the tool call
00:03:18that actually happened and we can see the state changes plus here's the recording which i wanted in
00:03:24the first place and that is what i want as a dev not just the bot worked i want to know why it worked
00:03:31when it fails i want evidence of this actually happening so what is dogra dogra appears to give
00:03:37us three different things out of all this we get a voice agent a visual workflow builder in the platform
00:03:44layer you usually have to build yourself the voice engine is the part that connects the caller the phone
00:03:50provider speech to text the llm and text to speech that is what makes the call actually happen the
00:03:57workflow builder is where you design the logic of this whole system so instead of hard coding every
00:04:03prompt branch api call and transfer you can map out the flow visually so huge win here i like these kind
00:04:09of maps ask this question wait for the answer that's kind of what we're mapping out here i can call this
00:04:15api branch here transfer there that kind of logic should be easy to change then to all this there's
00:04:21the platform layer testing tracing recordings analytics that is the boring stuff every series voice project
00:04:28eventually needs with all this you can bring your own providers your own llm and your own tts
00:04:34because dogra is open source you can inspect the code change how it works and self-host it as of this recording
00:04:41github stars are low so this is a super new find that i found but it's honestly a rather cool one
00:04:47now let's compare dogra to other things we already have out here you have three main ways to build
00:04:51voice agents first is hosted platforms vappy bland retail these are good when you want to move fast and
00:04:58you don't want to run infrastructure you get clean dashboards apis transcript testing tools all that's
00:05:04really useful but you start to lose control right there if the platform changes pricings you deal with
00:05:10it if the platform changes limits deal with it right if you need custom deployment anything like that
00:05:17again you might hit a wall hosted tools are fast though so i guess that's a win you have some of these
00:05:23raw frameworks like uh i came across pipe cap though code a live kit i think is one of them
00:05:30these give you a lot more control you can build almost anything but now you're building everything
00:05:36around this framework off ui workflow editor so that's a big trade-off using things like that
00:05:42now dogra is still way too new but it's here and i think their bet is kind of simple what if you could
00:05:49use a visual voice agent builder without giving up the self-hosting choosing a provider tracing and
00:05:56control that's what this appears to be write code where code matters use the builder where your flow
00:06:02matters inspect the runtime when things break and swap providers when costs change self-hosting gives
00:06:09us a lot of control which is huge vappy bland retail are best for fast hosted deployment but the trade-off
00:06:16cost locking and less control if you enjoy coding tools like this be sure to subscribe to the better
00:06:22stack channel we'll see you in another video

Key Takeaway

Dograh provides a self-hosted, open-source alternative to voice AI platforms, combining visual workflow orchestration with the transparency needed to debug complex call failures.

Highlights

  • Dograh is an open-source alternative to proprietary voice AI platforms like Vapi, allowing developers to self-host, inspect, and control their own agent infrastructure.

  • The platform provides a visual workflow builder that manages prompts, API tool calls, and call transfers without requiring custom orchestration code.

  • Developers can deploy Dograh locally by running a single docker-compose up command.

  • Testing and debugging tools built into the platform provide transcripts, traces, state changes, and audio recordings for every call.

  • Users can bring their own voice and LLM providers to the platform, avoiding vendor lock-in and platform-specific pricing models.

  • The tool separates low-level voice engine connectivity from high-level business logic design via a canvas-based interface.

Timeline

The Problem with Hosted Voice AI

  • Hosted voice AI platforms impose platform fees on top of LLM and telephony costs while limiting user control over the system.
  • Voice agents require complex orchestration of speech-to-text, text-to-speech, LLMs, and API integrations, making failure difficult to diagnose on closed systems.

Building a voice agent involves multiple moving parts that often fail in non-obvious ways during messy, real-world calls. When a call fails on a hosted platform, identifying whether the error originated from the prompt, the model, or the integration layer is often impossible.

Building Agents with Dograh

  • Deployment occurs via Docker, simplifying the setup of the entire agent infrastructure.
  • The interface functions as a developer-focused visual canvas where users define prompts, API tools, and call routing logic.
  • The platform supports advanced interactions like CRM lead qualification and automated transfers to human agents.

The setup process requires cloning a repository and running a docker-compose command. Within the UI, developers map out agent behavior visually, replacing manual orchestration code with a series of nodes for prompts, decision branching, and external API tool calls.

Debugging and Architecture

  • The platform layer offers built-in observability including call transcripts, trace logs, state change history, and audio recordings.
  • Dograh provides three distinct layers: a voice engine, a visual workflow builder, and an observability platform.
  • The architecture allows users to swap out LLM, STT, and TTS providers to manage costs and technical requirements.

Dograh addresses the need for evidence-based debugging by recording every aspect of a call interaction. By keeping the voice engine and workflow logic separate from the LLM and telephony providers, the system gives developers the control to change infrastructure without rebuilding their entire application logic.

Community Posts

No posts yet. Be the first to write about this video!

Write about this video