Myra Agent
Myra Agent is the chat assistant built into your dashboard. Ask for something in ordinary words — build me an agent, call these leads, how did last week go — and she does it, showing each step as she goes and asking only for what she genuinely needs.
Find it at Dashboard → Myra Agent.
What Myra is
Everything you can do by clicking through the dashboard, you can do by asking Myra instead. She is not a help bot that points you at a page — she carries the job out: creating the assistant, placing the call, connecting the sheet, pulling the numbers.
The difference matters most for the fiddly jobs. Setting up a tool that talks to your own API normally means knowing your endpoint, your authentication and your parameters before you start. Asked in chat, you paste what you have and Myra works out the rest, asking once for anything still missing.
Where to find her
The full page
Dashboard → Myra Agent. The whole window, with your past conversations in a sidebar on the left. Best for real work.
The floating button
A small Myra button sits on every dashboard page. Open it for a quick question without losing the screen you are on.
Your first chat
A new chat opens with a few examples you can tap to send. They are a good way to see the shape of a request before you write your own.
“Which of my assistants get the most calls?”
See which agents are busiest
“Summarise my last 5 customer calls”
Recap recent conversations
“Create a sales agent for my website”
Spin up a new voice agent
“Call these leads tomorrow at 10am”
Schedule an outbound campaign
The dot under the Myra Agent title shows the connection: green when she is ready, yellow while connecting. If it goes red, a Reconnect button appears next to it.
What you can ask
You do not need to phrase things the way this page does — these are just the areas Myra covers, with real examples of how people ask.
Try saying
- “Create a receptionist agent for example.com”
- “Make my sales agent friendlier and stop it repeating the price twice”
- “Switch the Skyline Properties agent to a faster model and make it speak Tamil”
- “Change the welcome message on my support agent”
Try saying
- “Call 98xxxxxx10 as a test”
- “Call everyone in this sheet one by one”
- “Call these numbers: 98xxxxxx01, 98xxxxxx02, 98xxxxxx03”
- “End the call that is running right now”
Try saying
- “Run a campaign on these 40 leads with my follow-up agent”
- “Pause the campaign I started this morning”
- “Drip-call my connected sheet every 5 minutes starting at 3pm”
- “Send me a call summary every Monday at 9am”
Try saying
- “Connect this Google Sheet to my sales agent”
- “Let my agent book meetings on my calendar”
- “Build a tool that checks order status on our API”
- “Connect the Linear MCP server”
Try saying
- “Read this sheet and tell me how many leads have no phone number”
- “Add a Status column and mark everyone I called today”
- “Pull the numbers out of this PDF and call them”
Try saying
- “How did my calls go last week?”
- “Which of my assistants has the best connect rate?”
- “Which leads from yesterday sound hot?”
- “Show me the transcript of my last call”
Try saying
- “Make me an agent that handles calls like this recording”
- “Improve my support agent using this call”
- “What was said on this recording?”
Try saying
- “What does this company do? https://example.com”
- “How do I get a phone number verified in India?”
- “What do I need to connect Zoho?”
How a task runs
Bigger jobs — creating an assistant, connecting a sheet, building a tool — run as a guided sequence rather than a single reply.
You see the steps as they happen
Myra shows what she is doing while she does it — reading a sheet, checking your numbers, creating the assistant. Finished groups collapse so the conversation stays readable.
She asks only what she cannot work out
Anything already in the conversation — a number you mentioned earlier, a file you attached, the assistant you have been discussing — is used without asking again. Questions come as buttons where there is a fixed set of answers, and there is usually an option to type your own.
Credentials go in a secure box
When something needs an API key or token, the input switches to a masked field. The value is never shown back in the conversation.
Permission cards stay until you act
When a task needs you to grant access — signing in to Google, for example — a card with a button appears and stays put, even as the conversation moves on, so you can come back to it.
You confirm before anything irreversible
Starting a campaign, dialling a list, or deleting something shows one confirmation with the details first. Reading, listing and reporting happen straight away.
Working with files
Use the paperclip in the message box, or drag a file straight onto the chat. Attach it with a message saying what to do with it — “call everyone in this list”, “which of these rows has no email?”.
| Type | Formats |
|---|---|
| Spreadsheets | CSV, TSV, XLS, XLSX, ODS |
| Documents | PDF, DOCX, TXT, MD, HTML, XML, JSON, LOG |
| Call recordings | MP3, WAV, M4A, MP4, OGG, OPUS, WEBM, FLAC, AAC, AMR |
Up to 20 MB per file
Attach more than one if you need to. Files upload as you pick them; the send button waits until they are ready. A file that is too large or of an unsupported type is flagged immediately, before you send.
You can also just paste a link. A Google Sheet, Doc or Drive link works the same way as an upload — Myra asks for access the first time and remembers it afterwards.
Staying in control
Pause and resume
Pause a long job and pick it up later from the same step. Resume from the button, or with ⌘/Ctrl + Enter. Typing something new is also a valid way out of a pause — Myra moves on to the new request.
Stop
Stops the current reply or job where it is. Press Esc or the stop button. Whatever was already said stays in the conversation, marked as stopped.
Edit what you sent
Hover a message you sent and rewrite it. The conversation continues from that point and the replies below it are replaced — useful when you gave the wrong number or the wrong assistant name.
Start fresh
New Chat gives Myra a clean slate. Worth doing when you switch to a completely different job — a long conversation about one thing makes the next thing slower to get to.
Chats and history
- Every conversation is saved and listed in the left sidebar, newest first, with the date. Click one to pick it up exactly where you left it.
- Delete a conversation from the sidebar when you no longer need it.
- Collapse the sidebar with the arrow to give the conversation the full width.
- Each chat keeps its own context. Myra remembers what you told her within a conversation, so you can say “call that number again” without repeating it.
Limits and connection
Conversation length
A single chat can only run so long before replies slow down. When it reaches the limit, the message box tells you and you start a new chat. Everything already said stays in the sidebar.
Usage limit
Very heavy use can hit a temporary cap. The message box shows a countdown, and sending unlocks itself when the window resets — there is nothing to do but wait.
If she goes offline
A red dot and a Reconnect button appear in the header. Press Reconnect; your conversation is still there.
Getting better answers
Say the numbers, don’t make her ask
Paste phone numbers straight into the chat in any format — one per line, comma-separated, or in a sentence. Myra reads them from the message and will not ask you to upload a file you don’t need.
Name the assistant when you have several
“Update the Skyline Properties agent” is faster than “update my agent”, which costs a round of questions when there is more than one.
Say when, if it matters
Scheduling only happens when you ask for it with a time: “tomorrow at 10”, “every Monday at 9”. Without a time, Myra does the job now.
One job at a time
Myra works through a request in order. If you need two unrelated things, ask for the second once the first is done — or start a new chat for it.
Correct her in plain words
“No, use the other number”, “make it shorter”, “skip that step” all work. There is no command syntax to learn.
Common questions
Will Myra call people or spend money without asking?
No. Anything that dials, charges or deletes shows a confirmation with the details first. Reading, listing and reporting happen without one.
Can she see all my data?
She works within your own account — your assistants, your numbers, your calls, and any file, sheet or document you point her at. She only reaches an outside service when you ask her to connect one, and only with the access you grant at that point.
What happens to an API key I paste?
It is collected in a masked field and stored with the connection it belongs to. It is never printed back into the conversation.
Do the calls she places show up normally?
Yes. Everything appears under Call Logs with the recording, transcript, duration and outcome, exactly like a call started from the dashboard.
Can I keep working while she runs a long job?
Yes. The pill in the header shows Working or Paused wherever you scroll, and you can pause the job and come back to it.
She says the conversation reached its limit — what now?
Start a new chat. Long conversations are capped so replies stay fast. Your earlier chats stay in the sidebar and nothing is lost.
Is this the same as the chat widget on my website?
No. Myra Agent runs your Vomyra account from inside the dashboard. The embeddable widget is a separate feature that talks to your customers.
Related pages
Train My Agent
Teach an assistant to talk like your best calls, from real recordings.
Outbound Campaigns
Run a lead list as a managed campaign you can track and pause.
Google Sheets Integration
Connect a sheet to an assistant for lead capture and call records.
Connect apps with MCP
Give your assistants tools from Linear, Notion, Slack and more.