Flowyte is API-first. Every action you can take in the dashboard — creating an agent, giving it knowledge, adding a skill, publishing it, and testing it — is also a single HTTP call against the Flowyte API. That means you can build and operate a phone agent three ways: by hand in the dashboard, by code against the API, or by handing the docs to an AI coding agent and asking it to set things up for you.
The video above shows that last path end to end: a developer points an AI coding agent at the Flowyte docs and asks it to stand up a working receptionist for a new dental office. The agent reads the docs, mints a key, creates the agent, configures it, adds knowledge and a skill, publishes it, and runs a test call — all over the API. The written walkthrough below covers the same sequence so you can follow it yourself at your own pace, whether you are writing the calls by hand or driving them with an AI assistant.
Who This Is For
This guide is for developers, automation builders, and AI coding agents — anything that can read documentation and make HTTP requests. You do not need the dashboard open to build a Flowyte agent. If you would rather click than code, the visual builder does the same things through guided forms; the API is for when you want to script it, integrate it into your own tooling, or let an AI agent do the setup.
Why an API-First Phone Agent Matters
A traditional phone system is something you configure once, by hand, in a console. An API-first one is programmable: you can version it, automate it, reproduce it across accounts, and wire it into the rest of your stack. The same /api/v1 that the Flowyte dashboard uses is available to you directly, so there is nothing the UI can do that your code cannot.
It also unlocks a newer workflow. Because everything is a documented HTTP call, an AI coding agent can read the docs and operate Flowyte on your behalf. You describe the outcome — "set up a receptionist for my dental office" — and the agent translates that into the right sequence of API calls.
Everything the dashboard does, the API does. If you can see it in the builder, there is an endpoint for it. That makes the API a safe place to automate: you are not working around the product, you are using the same surface it uses.
Before You Start: The Docs
Everything you need lives at docs.flowyte.com. A few resources are worth knowing about up front, especially if you plan to hand the work to an AI coding agent:
- A "For AI agents" guide that explains, end to end, how an LLM-based coding tool builds and operates a Flowyte agent over the API.
- Machine-readable docs: a plain-text index at
/llms.txt, the full docs as a single file at/llms-full.txt, and the Markdown of any page by appending.mdto its URL. - The OpenAPI contract, available from the API Reference, so a tool can discover every endpoint and its shape.
The base URL for every call is https://builder.flowyte.com/api/v1. Authentication is a secret API key sent as a bearer token: Authorization: Bearer YOUR_FLOWYTE_API_KEY.
Step-by-Step: Build an Agent Over the API
Point yourself (or your AI agent) at the docs
Open the docs at docs.flowyte.com, or give the URL to an AI coding agent and ask it to read the "For AI agents" guide and the OpenAPI contract. Everything a person can do in the dashboard is available over the API, and the docs map each action to an endpoint. If you are driving this with an AI assistant, this is the only step you do by hand — the rest it can run for you.
Mint a scoped secret key
In the dashboard, go to Developer and create an API key. Give it a clear name like brightside-dental-setup. Choose an environment — Test keys are isolated from live traffic and billing, so use Test while you build. Then select only the scopes you need: granular permissions like read agents, manage agents, manage knowledge, and manage skills. The key can do exactly what its scopes allow and nothing more. You will see the secret once, so copy it immediately.
Create the agent
Make a single POST to create the agent. Capture the id from the response — that is your agentId for every call that follows.
curl -X POST https://builder.flowyte.com/api/v1/agents \
-H "Authorization: Bearer YOUR_FLOWYTE_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "name": "Brightside Dental" }'
The response is a standard envelope; read data.id to get your new agent's ID. A freshly created agent starts in a draft state.
Configure persona, hours, and routing
Send a PATCH to /agents/:id to shape how the agent behaves: its greeting, its goals, its never-do guardrails, its business hours and timezone, and what to do in an emergency — for a dental office, that might mean routing urgent calls to an on-call line. This is the same persona and routing you would otherwise fill in through the dashboard forms.
Add a knowledge source and wait for it to index
POST a knowledge source to /agents/:id/knowledge — for a clinic, that is your services, insurance, hours, and payment information. Indexing happens asynchronously, so poll the source until its status is indexed before you rely on it. A small knowledge base indexes into a handful of chunks in moments.
Adding knowledge is asynchronous. The call returns right away, but the source is not ready until it reports indexed. Poll the source's status rather than assuming it is live the instant you upload it.
Add a skill
POST a skill to /agents/:id/skills to give the agent a capability beyond answering questions. A common one is an "appointment request" skill that collects a caller's name, phone number, reason for the visit, and preferred time, then emails it to you. In the dashboard these are built with guided forms, so you can always see exactly what caller data leaves your system; over the API you send the equivalent configuration.
Publish the agent
POST to /agents/:id/publish to take the agent live. Publishing creates a version — v1 on your first publish — and makes the agent available on both voice and chat. Until you publish, your changes only affect the draft.
Test it, then go live
Before you point a real number at it, POST to /agents/:id/simulate to run a sample caller turn and read back the transcript. This is the API equivalent of the in-dashboard tester: you confirm the agent greets correctly, answers from your knowledge, and triggers the skill when it should. When the transcript looks right, assign a phone number and your agent is live.
Build with a Test key first. Because Test keys are isolated from live traffic and billing, you can run the whole create-configure-publish-simulate loop without touching production or incurring usage. Switch to a Live key only when you are ready to take real calls.
The Shape of Every Call
The pattern is consistent across the whole API, which is what makes it easy to script or hand to an AI agent:
- One base URL:
https://builder.flowyte.com/api/v1. - One auth header: a bearer token with a scoped secret key.
- One response envelope, so you always read your data from the same place.
- Asynchronous work (like indexing knowledge) reports a status you poll until it is ready.
Once you understand those four things, the rest is just choosing the right endpoint for the action you want — and the docs map every dashboard action to one.
Let an AI Agent Do the Setup
The most hands-off path is to skip writing the calls yourself. Point any AI coding agent — an LLM-based tool that can read docs and make HTTP requests — at docs.flowyte.com, hand it a Test API key, and describe what you want in plain English. Because the docs are machine-readable and the API mirrors the dashboard exactly, the agent can run the full sequence above on its own: create, configure, add knowledge, add a skill, publish, and simulate. Flowyte also includes an in-dashboard build copilot that configures an account from plain English using the same API, if you would rather stay inside the product.
What's Next
- Register webhooks in Developer so your systems get notified when calls or events happen.
- Script repeatable setups so you can stand up a new agent for each location or client with one run.
- Move from Test to Live once a simulated agent behaves the way you want.
- Compare approaches by reading how the same agent is built in the visual builder, or why teams replace a legacy IVR with a conversational agent.
Pricing is usage-based — prepaid credits where 1 credit equals $0.01, voice from $0.11 per minute on pay-as-you-go, and phone numbers at $2 per number per month. See the pricing page for current plans and full details.
Start Building on the Flowyte API
Mint a scoped key, create an agent, and publish it with a handful of HTTP calls — or hand the docs to an AI coding agent and let it build for you.
Start Building FreeCommon Questions
What is the Flowyte API?
The Flowyte API is the programmable interface behind the Flowyte dashboard. It is API-first, which means every action you can take in the dashboard — creating an agent, adding knowledge, adding a skill, publishing, and testing — is also a single HTTP call. You can build and operate AI phone agents entirely over the API.
What is the base URL and how do I authenticate?
Every request goes to the base URL https://builder.flowyte.com/api/v1. You authenticate with a secret API key sent as a bearer token in the Authorization header, in the form Authorization: Bearer YOUR_FLOWYTE_API_KEY. The key never goes in the URL.
Can an AI coding agent set up Flowyte for me?
Yes. The docs are machine-readable and the API mirrors the dashboard exactly, so any AI coding agent that can read documentation and make HTTP requests can run the full setup for you. You point it at the docs, give it a Test API key, and describe what you want, and it creates, configures, publishes, and tests the agent. Flowyte also includes an in-dashboard copilot that configures an account from plain English using the same API.
How do I get an API key and choose scopes?
In the dashboard, open Developer and create a key. You give it a name, pick an environment of Test or Live, and select granular scopes such as read agents, manage agents, manage knowledge, and manage skills. The key can do exactly what its scopes allow and nothing more. The secret is shown only once, so copy it immediately, and if you lose it you rotate the key.
Is the API the same as the dashboard?
Yes. The dashboard uses the same /api/v1 that is available to you. Anything you can do in the dashboard you can do over the API, so working through code is not a workaround — it is the same surface the product itself uses.
What can I build or automate with the API?
You can create an agent, configure its persona, hours, and routing, add knowledge sources, add skills like collecting an appointment request and emailing it, publish to voice and chat, and run test simulations. Because it is all scriptable, you can also register webhooks and stand up repeatable setups for new locations or clients in a single run.
What is the difference between Test and Live keys?
Test keys are isolated from live traffic and billing, so you can build and run the full create-configure-publish-simulate loop without touching production or incurring usage. Live keys take real calls and count toward billing. The recommendation is to build with a Test key first and switch to a Live key only when you are ready to go live.
How does adding knowledge work over the API?
You POST a knowledge source to the agent, and indexing happens asynchronously. The call returns right away, but the source is not ready until its status reports as indexed, so you poll the source until it is done before relying on it. A small knowledge base indexes into a handful of chunks quickly.
How much does it cost to use Flowyte?
Flowyte uses prepaid credits where 1 credit equals $0.01. Voice starts at $0.11 per minute on pay-as-you-go and drops on higher plans, and phone numbers are $2 per number per month. Test keys are isolated from billing, so building and simulating with a Test key does not incur usage. See the pricing page for current plans and full details.
About the Author
Flowyte Support
Support Team
Helping businesses automate phone calls with AI. Questions? Reach us at support@flowyte.com.
