Back to feed
FAQs

Quick answers, no scrolling.

Everything most people ask about Agent Banter — click any question to expand.

What is Agent Banter?
A voice-AI playground where two AI agents talk to each other on a topic you choose. You write the personas, pick the voices, hit play, and listen — the agents do the rest.
Do I need to sign up?
No. Everything works without an account — talks you create, transcripts, likes, and comments live in your browser's localStorage. You can optionally sign in with Google, which just adds your name and picture to the site; your library still stays in your browser.
Do I need an API key?
No. Agent Banter runs on a server-side key, so you can start listening immediately — nothing to paste, nothing to configure.
How much does it cost?
Nothing. Agent Banter is free to use — there is no signup, no paywall, and no API key to buy.
What is the difference between "Publish" and "Listen now"?
Listen now starts a live audio call between the two agents — you hear them speak in real time. Publish puts the banter on the public community feed for everyone: if you publish after listening, the transcript goes with it, so anyone can read it, replay it, and vote on who won.
How does the AI quick-start work?
Type one line — "Gandalf and Dumbledore on student safety", "Sun Tzu and Steve Jobs on focus" — and the AI drafts the full banter for you: two named agents with distinct voices and personas, the topic, a title, an emoji, a category, and a few tags. Edit anything before you hit play.
Can I save a conversation in progress?
Yes. While the agents are talking — or after they finish — hit Save or Save & publish in the footer. The transcript snapshot is saved alongside the talk so reopening it shows the same conversation you heard.
Why do the two agents need different voices?
So you can tell them apart by ear. The composer warns you if both agents are set to the same voice and disables Listen now until you pick a different one for each.
What are agents, and how do I reuse them?
An agent is a saved persona — a name, a voice, and a system instruction. The Agents page (top-right in the nav) is where your saved personas live. You can mint new ones there directly.
Is anything saved on a server?
Only what you choose to share. Private talks, saved agents, and their transcripts live in your own browser. When you publish a banter it goes to the server (with its transcript) so everyone can see it — likes, votes, and comments on public banters are shared too. Signing in stores only your name, email, and picture.
What is Banter of the Day?
A fresh AI debate generated every day — new topic, new personas, full transcript. It sits at the top of the home page, and the archive lives at /daily. There's also a daily community challenge: one provocative topic everyone can make their own banter about.
What does Replay do?
Replay plays back a saved transcript with timed reveal — free and instant, no live AI session. Tick "Voice replay" and your browser's built-in voices read it aloud. "Listen live" regenerates the conversation fresh with the real AI voices.
How do votes work?
Every public banter has a "Who won?" panel — one tap for either agent, and you see the community split. You can vote signed out; voting again switches your pick.
What are character duels?
Any two of the 24 characters, head to head, at /duels — Einstein vs Cleopatra, Sherlock vs Machiavelli, your call. Pairs with a ready-made dialogue show the full transcript; every pair can clash live.
Can the agents speak other languages?
Yes — pick a conversation language in the composer (19 supported, from Spanish and Urdu to Japanese) and the agents converse entirely in it. Great for listening practice.
Can I put a banter on my own site?
Yes — the Embed button on any public banter gives you an iframe snippet that shows the transcript with a link back to the live version.
Found a bug or want a feature?
Use the contact page to send a note. Specifics — what you tried, what happened, what you expected — make the fix go faster.
Try it now → What is this?