Vogent

What Vogent is

Vogent is a platform for building voice agents on real phone numbers. Its homepage calls it "the all-in-one platform for building humanlike, intelligent, and effective voice agents", under the headline "Build AI Voice Agents. Fast." It has two front doors. The site's navigation separates a No-Code path from a For Developers path, so the same product serves operations teams that want a visual builder and engineers who want to work with the API underneath. The legal name in the footer is Monoid, Inc., doing business as Vogent. The pages do not say where the company is based.

The company says it is "Powering millions of phone calls for thousands of organizations" and reports "1M+ calls made per month". It has also launched a separate product, Vogent Voicelab, introduced on the homepage as ultra-realistic text-to-speech that lets you run top text-to-speech models.

What you build and how

The Flow Builder. The no-code editor is introduced with "Drag. Drop. Talk." and the promise that you can "create intelligent and effective agents without a single line of code". You piece together agents from "intelligent building blocks": blocks that ask questions and probe for answers, hold conversations on specified topics, and call external tools when necessary. The company says you can go "from idea to agent in minutes".

Phone numbers. Agents are hosted on live phone numbers. You select a number by area code and use it to host your agent.

Models built for calls. The homepage says Vogent has "built multiple layers of models, covering everything from IVR and voice activity detection to custom voices and conversational flow routing". There are two that stand out. The first is IVR detection and navigation, "to help you reach a human every time". The homepage shows an automated menu offering claims status on 1, and the agent's response of pressing 1 toward the goal of reaching a representative. The second is phone-trained language models. In addition to off-the-shelf models like GPT, you can use "Vogent's custom-built conversational models that have been fine-tuned on millions of noisy phone calls". The site shows a mistranscription being corrected: "The appointment is for James Wright?" becomes "The appointment is for James Wright, right?"

Tools and knowledge. The features section lists "Functions and Tools", which let an agent hit APIs and query knowledge bases live on the call, then use what it gathers to compose its responses. Knowledge bases let you upload large documents and have the agent look up answers in real time.

Call transfers. An agent can hand a call off to a human at the right time, or have multiple AI agents work together.

Reviewing calls. The call history view shows every call recording and transcript. You can "analyze how decisions were made, and test new agents against old calls". A related feature, Counterfactual Analysis, evaluates new agents on past conversations "seeing how they would have handled different scenarios". The homepage shows an outcome extraction view that pulls structured fields such as patient name, appointment date and appointment time out of a call.

Post-call workflows. You can trigger integrations and workflows after every call, tying dial outcomes to actions "so you can update your systems or follow-up with callers based on parsed call outcomes".

For developers. The developer section is titled "Developer? Access everything under the hood." It offers a rich API and native SDKs with primitives for turn taking, conference calling, IVR detection and custom models, described as "the full voice stack" for building a vertical voice AI product. You can bring your own models by entering the URL of an OpenAI-compatible API: "Give it a voice with a few lines of code."

Detection and turn handling. Voice activity detection is named alongside IVR detection among the layers of models, and the developer primitives include turn taking and conference calling. For a builder, that means the hard parts of a phone call, knowing when a caller has finished speaking, getting through a phone menu and bringing a third party onto a call, are listed as platform features rather than things you wire up yourself.

Learning from past calls. The call history and Counterfactual Analysis features work together. Because every recording and transcript is kept, a revised agent can be run against earlier conversations before it handles a live one, which gives a team a way to check a change without dialing real customers. The homepage presents this as part of maintaining and refining an agent, next to "Dozens of tools to build, maintain, and refine your voice agent".

Community and docs. The navigation links to docs, a blog and a Discord community, which are the places a developer would look for setup detail that the homepage leaves out.

Vogent is a voice product. The pages mention no text chat widget or messaging channels, so the chatbot-builders reader should read it as a way to build conversational agents for phone calls.

What the pages say about the product

The homepage gives three headline numbers. It cites "200ms average response latency using Vogent's custom models and voices", a "75% average reduction in call center or labor spend" and "1M+ calls made per month". These are the company's own figures and the page does not explain how they were measured.

On security, the homepage says Vogent "offers HIPAA-compliant voices and workspaces, and is pending a SOC 2 Type II audit report." That is a specific statement: HIPAA is claimed, and SOC 2 Type II is not yet complete. The enterprise section adds custom-trained agents, dedicated high-volume infrastructure and support, and HIPAA-compliant workspaces. Fine-tuning, where you "work with Vogent's solutions engineers to guardrail your agents", is "available on the enterprise plan".

The pages publish no prices or plan limits. The site has a Pricing link and an FAQ whose questions include whether there is a free trial and whether you can get help building an agent or use pre-built templates, but the answers are not on the page. Anything about cost, trials or templates needs to be confirmed with the company.

The site does not name customers or give case studies on this page, though its menu points to a Customers section.

Who it suits

Vogent suits teams that make or take structured phone calls at volume and want to build the agent themselves. The IVR navigation model and phone-trained language models are aimed at calls that run into automated phone menus and noisy real-world conversations. The Flow Builder lets a non-engineer assemble the conversation, while the API and SDKs give a developer a way to embed voice in a product. The ability to bring your own OpenAI-compatible model is a real escape hatch for technical teams.

It is a weaker fit if you need published pricing to plan a budget, since the pages show none. A team that requires a completed SOC 2 Type II report should note that the company describes it as pending. And anyone looking for a website chatbot builder will find the pages focused on phone calls.