Visual workflow builder
Design the conversation, not the code.
Drag agent nodes onto a canvas, branch with conditional edges, add global nodes for questions that can come up anywhere, and pull variables out of every call.
Your phone line, run by agents you design.
A visual workflow builder for voice AI that makes and takes real calls — bring any speech or language model, or host the whole stack yourself.
Web call00:42
- AgentHappy to get you booked. What day works best?
- CallerThursday, ideally before lunch.
- AgentThursday morning it is. What name should I put this under?
Captured
preferred_time: Thursday, morningRuns the models you already trust
Mix 25+ speech, language and voice providers, or run open models on your own hardware.
- OpenAI
- Anthropic
- Deepgram
- ElevenLabs
- Groq
Design, run and review every call in one place
Nexon covers the whole life of a phone agent: building it, testing it, putting it on a real number, and checking its work afterward.
Inbound and outbound calling
Answer calls and place them, at scale.
Connect Twilio or Plivo numbers for inbound, or upload a CSV to run an outbound campaign with automatic retries and voicemail detection.
Test before you ship
Talk to your agent before your customers do.
Start a web call from your browser while you build, and use test mode to try changes without touching production data.
Built-in QA
See exactly what happened on every call.
Every run keeps its recording, transcript and extracted variables, and a QA node reviews prompt quality across your whole workflow.
Bring any model
Pick your models, and your costs.
Combine speech-to-text, LLM and text-to-speech from 25+ providers, use realtime speech-to-speech, or run open models like Kokoro TTS and Voxtral STT on your own servers with Ollama.
Developer-friendly
Wire calls into the rest of your stack.
Use the REST API with Python and Node SDKs, trigger agents from your own systems, receive webhooks, and ground answers in a knowledge base.
Also included on every plan: organizations with owner, admin and member roles, member invites, per-organization API keys, and prebuilt templates to start from.
From blank canvas to live calls in three steps
All three happen in the same workspace, so the agent you test is the agent you launch.
-
Build
Start from a template or a blank canvas. Write each agent's instructions, connect them with conditions, and choose the details to capture.
-
Test
Open a web call and talk through the flow yourself. Test mode keeps these conversations out of your production data.
-
Launch
Point a Twilio or Plivo number at your workflow for inbound calls, or upload a contact list to start an outbound campaign.
Start with the calls you already make
Each of these begins as a prebuilt template you can edit on the canvas, so your first agent starts from a working flow instead of an empty page.
Appointment booking
Pick up every booking call, collect the details, and confirm the visit, including after hours.
Example line: “Is this for a new patient visit or a follow-up?”
Lead qualification
Call new leads while they're still interested, ask your qualifying questions, and log the answers.
Example line: “Roughly how many locations are you looking to cover?”
Customer support
Answer common questions from your knowledge base at any hour, and capture the details for anything that needs a person.
Example line: “Our return window is 30 days. Do you have your order number handy?”
Payment reminders
Remind customers about upcoming or overdue balances with a polite call and a clear next step.
Example line: “I'm calling about the balance due Friday. Is now a good time?”
Pricing
Coming soon...
We're still working out our plans. Start free in the meantime, or book a demo and we'll talk through what your call volume would cost.
Common questions
Still deciding? Book a demo and we'll walk through your use case with you.
What counts as a minute?
A minute is counted while a call is connected, inbound or outbound, from the moment it's answered until it ends. Building, editing and storing workflows is free, so an idle workflow costs nothing.
Each plan includes a set number of minutes. Past those, extra minutes are billed per minute.
Can I bring my own API keys?
Yes. Connect your own accounts for OpenAI, Anthropic, Deepgram, ElevenLabs and the other supported providers, and choose a different provider for speech-to-text, the language model and text-to-speech if you like. You can also point Nexon at open models you host yourself.
Is my data isolated from other customers?
Yes. Every organization's data is scoped and isolated. Workflows, recordings, transcripts, extracted variables and API keys belong to one organization and are never visible to another.
Inside your organization, owner, admin and member roles control who can do what, and test mode keeps test calls separate from production data.
Can I self-host?
Yes. Nexon is self-hostable, so you can run the platform on your own infrastructure. Pair it with self-hosted models such as Kokoro TTS and Voxtral STT, or LLMs served through Ollama, to keep speech and language processing on hardware you control.
How do concurrent calls work?
Concurrent calls are the number of calls your organization can have live at the same moment, inbound and outbound combined. Each plan sets its own limit.
When an outbound campaign has more contacts than open lines, Nexon dials the next contact as soon as a line frees up. Need more lines? Book a demo and we'll set concurrency to match your volume.
Put an agent on your phone line this week.
Start free, build your first workflow from a template, and call it from your browser before you connect a single number.