Building an agent by voice Guide

Praxivara's AI builder can set up an agent from a plain-English conversation, and you don't have to type any of it. Voice mode lets you describe what you want out loud and have the builder wire up the agent while you keep your hands free.

When voice mode is a good fit

Building by voice works the same way as building by text — you're talking to the AI builder on the agent's Build tab, and it configures the agent as you go. Voice is especially handy when you're away from your keyboard, thinking through an idea, or want to describe a workflow the way you'd explain it to a colleague.

Everything the builder sets up by voice appears on the agent's Configure tabs just as it would if you'd typed it, so you can always review the result on screen afterward.

Start a voice session

1. Open the agent you want to work on, or create a new one, and go to its Build tab.
2. Start voice mode from the build conversation, then allow microphone access if your browser asks for it.
3. Wait for the builder to signal that it's listening, then start talking.

Tip: Voice mode is a natural conversation. You don't need special commands or exact phrases — describe the job the way you'd explain it to a new team member, and the builder will ask follow-up questions if it needs more detail.

Describe what you want the agent to do

Speak in outcomes, not settings. Tell the builder what the agent is for, what it should produce, and how you want to be kept informed. For example:

• "Build an agent that watches for new leads and drafts a personalized reply for each one."
• "I want it to summarize my Slack channels every morning and email me the highlights."
• "Have it write a weekly SEO blog post and save the draft for me to review."

As you talk, the builder can add the tools and integrations the agent needs, install the right skills (proven, step-by-step playbooks) for the job, and set up how the agent notifies you. You can ask for changes at any point — "also let it send me a WhatsApp message when it's done" — and the builder will adjust.

Add capabilities as you go

You can grow the agent's abilities in the same conversation. A few things you can ask for by voice:

• Tools and integrations: "Give it access to Gmail" or "let it search the web." If an integration needs a connected account, the builder will let you know so you can connect it.
• Skills: "Add the brand-monitoring skill." Installing a skill also brings along the extra tools that skill relies on.
• Notifications: "Email me a summary after each run," or "keep it quiet unless something needs my attention."

Some setup steps are best finished on screen. Adding a secret value like an API key, uploading a large knowledge file, or connecting an integration account are actions you'll complete with your hands — the builder will point you to the right place when one comes up.

Review and confirm

When you're happy with what you've described, take a moment to check the result:

1. Open the Configure tabs — Tools, Skills, Files, Secrets, and Knowledge — to see exactly what the builder set up.
2. Run a test to watch the agent work end-to-end before you rely on it.
3. If something isn't right, keep talking (or switch back to typing) and ask the builder to adjust. Version history lets you roll back to an earlier setup if you change your mind.

Tips for smoother voice sessions

• Give one clear instruction at a time, then let the builder respond before adding the next.
• Confirm names and specifics out loud — the account an integration should use, or the title of a report — so the agent is grounded in your details.
• You can move between voice and text freely; anything you set up in one carries into the other.

Building by voice is just a faster, hands-free way to have the same conversation. Describe the outcome you want, review what the builder set up, and confirm it with a test run — the agent is ready to go from there.