Docs

Plonom documentation

Everything you need to build, test and run voice and message agents.

Getting started

Plonom runs AI agents that answer your phone line and WhatsApp. You build and manage everything from the panel.

Create an account on the sign-up page, or sign in with your email and password.
Create an agent under Build → Agents.
Test it in the browser by voice or by message.
Publish it, then connect a phone number or your WhatsApp Business number.
Follow calls, chats and contacts under Data, and results under Monitor.

Your workspace name, industry and hand-off hours are in Settings → Business.

Create an agent

There are two kinds of agents. Voice agents talk on the phone and in the browser. Message agents reply in writing on WhatsApp and can be tested in the browser.

Go to Agents and click Create an agent, then choose Voice agent or Message agent.
Pick a starting point: Build from scratch, Start from a description, Your business info, or a template such as Front desk receptionist, Appointment booking, Lead qualification or Customer support.
The agent is created and the editor opens. Set the name, language and (for voice agents) the voice there.

Configure the agent

The editor has three columns: the prompt on the left, settings in the middle and the test panel on the right.

Model, voice and language: the AI model, the voice and the language the agent speaks.
Prompt: what the agent knows and how it behaves. Variables like {{customer_name}} are filled for every call.
Welcome message: the first sentence, or let the customer speak first.
Functions: end the conversation, transfer to a person and the integrations you switched on.
Knowledge base: the libraries the agent answers from.
Speech settings (voice): speed, stability, volume, responsiveness, interruptions, backchannel and background sound.
Call settings (voice): silence timeout, maximum length, reminders and voicemail.
Phone numbers (voice): which numbers this agent answers and calls from.
Chat settings (message): close the chat after silence and the closing message.
Post-call extraction: the fields saved after each conversation.
Data: how much of each conversation is stored, and which personal details are removed.

Test the agent

The test panel always uses the latest saved draft. Save before testing your changes.

Voice test: tap the circle and allow the microphone. Tap again to end.
Message test: type a message. Voice agents can be tested by message too, with the same prompt.
Test calls are billed like any other call and appear in Call history.

Publish and versions

Save draft keeps your changes without affecting live calls. Publish makes them live as a new version: the first publish is v1, the third is v3.

When you save after publishing, the badge shows Unpublished changes until you publish again. The Versions section lists every published version.

Knowledge base

A knowledge base is a library of documents, web pages and texts. Libraries live in the workspace; each agent chooses which ones it answers from.

Open Build → Knowledge base and create a library.
Add texts, web page addresses (one per line) or documents.
Wait until the status is Ready.
In the agent editor, open Knowledge base, tick the library and save.

Phone numbers

Bring your own number from Twilio, Telnyx, Vonage or your operator over SIP.

Open Deploy → Phone numbers and add a number.
Enter the number, the SIP termination address and, if your provider needs it, the SIP user name and password.
Choose the agent that answers incoming calls and the agent that calls out from this number.
Use Make an outbound call to place a real test call.

Batch calls

Call a list of contacts from one of your numbers. Each row of the CSV is one call.

Open Deploy → Batch call and click Create a batch call.
Upload the CSV, pick the number and the agent.
Send now, schedule it, or save it as a draft. Choose a calling window and how many calls run at the same time.

Every result is saved to the contact. You can also call a single contact back from the Contacts page.

Plonom sits between WhatsApp and your agent: it receives each message, saves the contact, keeps the chat history, asks the answering agent and sends the reply back on WhatsApp.

Connect your WhatsApp Business number on the WhatsApp page. Customers see your business name and logo.
Answering agent: choose which message agent replies, or the built-in assistant.
Missed call follow-up: when a call is missed, the agent can send the customer a WhatsApp message so the conversation continues there.

Integrations

Integrations work the same for voice and message agents. Connect them in System → Integrations, then switch each one on in the agent's Functions and save.

Save contacts (built in): the agent writes the customer's name, need, budget and a summary to Contacts during the conversation.
Cal.com: enter your API key and the event type ID. The agent checks free times and books the slot the customer picks.
MCP: enter the server address and, if needed, an access key. Its tools appear in the agent's Functions.
Zapier, Make, n8n: paste the webhook address. The agent sends each contact to your workflow.

Webhook payload sent to Zapier, Make and n8n (JSON):

{
  "name": "send_to_zapier",
  "args": {"name": "Omar Haddad", "phone": "+971505550101", "need": "2BR in Marina",
           "budget": "2M AED", "summary": "Wants a viewing on Saturday"},
  "call": { ... conversation details ... }
}

Developers can also use the REST API (Settings → API keys) and send events to their own address (Settings → Webhooks).

Calls, chats and contacts

Call history: every call with recording, transcript, summary, sentiment and outcome.
Chat history: every WhatsApp conversation, plus test chats if you show them.
Contacts: everyone your agents spoke to, with stage, need and source. Open a contact to read the conversation, change the stage, call back or take over.
Export: download calls or contacts as CSV.

Monitoring

Analytics: calls, chats and agent performance for the last 7, 30 or 90 days.
Live monitoring: calls happening right now across your agents.
AI quality: every finished call is checked; the ones with a negative sentiment, an unsuccessful outcome or a request for the team are listed for review.
Alerting: get an email or a webhook when a metric such as failed calls or negative sentiment crosses a threshold.

Billing

Your workspace has a plan that sets its limits, billed monthly or yearly (yearly is 10 months' price: 2 months free), and a balance in US dollars that pays for usage. Paid plans add their included voice minutes to the balance every month, also on yearly billing (Pro 300 minutes, Business 800); premium minutes and AI replies use the same allowance at their rates, and unused minutes do not carry over. Beyond that you top up by card from $15, with no upper limit; credit you top up does not expire.

Calls are billed per second from the balance; AI replies on WhatsApp are billed per message. The rates are on the Billing page and the pricing page.
Phone numbers (your own, connected over SIP) and knowledge bases cost nothing extra. If the balance stays below zero for 20 days, the workspace's phone numbers and knowledge bases are removed; admins get an email warning 5 days before.
Want us to set everything up? The Custom plan includes a done-for-you setup for a one-time $299 fee, which is not refundable.
New calls need a small positive balance. Incoming calls and WhatsApp keep going for a while below zero so customers are never cut off.
Every top-up and charge is listed under Transactions on the Billing page.

Settings and security

Business: workspace name, industry, who conversations are handed to and when, follow-up time and the WhatsApp number for alerts.
Team: invite members as Admin, Editor or Viewer.
Security: how long contacts are kept, sign out every session, and delete the workspace.
Audit log: every change with who made it and when.
My account: change your password and turn on two-step verification with an authenticator app.
Sign-in is locked for 15 minutes after 8 wrong attempts. Incoming events are signed and checked, keys stay on the server and every page is served over HTTPS.