Updated October 16, 2026. This is a documentation-only review of ZeroBot’s public interface, company update, terms and privacy policy. We did not create an account, buy a plan, test audio, or benchmark answers. Feature descriptions identify what the vendor presents; they are not performance results.
Image note: The retained hero is an archival product screenshot from this listing’s earlier marketing presentation. Its model names, layout and illustrated avatar are historical, not a current availability list or evidence of human-like conversation quality.
What is ZeroBot?
ZeroBot is an AI chatbot service operated by ZeroBot LLC, a New Jersey company identified in its terms of service. This listing concerns the product at zerobot.ai.
The public interface shows text chat and agent settings for instructions, avatars, voices and public or private visibility.
A March 2026 company update also describes an agent platform with separately priced actions, USDC payments on Base, and optional World ID verification. That expands the product beyond its earlier voice-chat presentation. Check an agent’s action charges separately when comparing the cost of a subscription.
Models and voice
The model menu lists OpenAI, Anthropic, Google, xAI and Groq. We did not verify voice quality, latency or plan-specific access.
For a useful comparison, choose one ordinary task and repeat the same prompt with the available options. Assess whether the answer follows instructions, acknowledges missing information and remains consistent across follow-up questions. A model name alone cannot establish which option will suit your task. Keep a separate note of response quality and usage costs rather than assuming the most expensive choice is automatically better.
Website embedding
The public page includes embed-code, appearance and allowed-domain controls. Its Realtime 2 voice option says a creator-supplied OpenAI key is required.
This supports a qualified website-embedding claim. We have not installed the widget or verified a production deployment. Before putting one on a business site, check mobile layout, accessibility, failure messages and a clear route to a person. Test whether it refuses to invent opening hours or policies when information is missing. Confirm who pays for visitor usage and any external provider charges.
Pricing and billing
Public page text lists minute packs of 20 minutes for USD $5, one hour for USD $12, and eight hours for USD $88. It also displays Pro at $100/month, without an explicit currency beside that label.
These are not verified checkout offers. Plus pricing, included credits, conversion to minutes, expiry, rollover and pack-to-plan relationships remain unverified. Before paying, check the purchase screen for currency, taxes, allowances, overage treatment and recurring charges.
The published terms say periodic subscriptions renew automatically. They allow cancellation through the auto-renew setting or by contacting [email protected], with access continuing through the paid billing period. Refunds are discretionary and, if approved, return to the original payment method; the terms also warn that a refund or chargeback may lead to account exclusion. Do not assume cancellation produces a refund. These are the vendor’s stated terms, not a determination of statutory consumer rights.
Privacy and limits
The privacy policy, dated March 25, 2025, says it collects chatbot conversations and uses information for purposes including model training. It names OpenAI, Google TTS and ElevenLabs TTS among third-party services. It says saved chat histories remain while you keep them saved, and describes rights to request access or deletion, subject to limitations. It does not establish a universal no-retention or no-training arrangement.
The terms also describe inputs as non-confidential and allow their use to improve AI systems. Avoid entering confidential client material, passwords or sensitive personal details. Before workplace adoption, ask the provider how the current models, voice processing, deletion and training controls apply to your account.
ZeroBot’s terms disclaim professional advice and guarantees of accuracy. This review makes no claim that an agent provides mental-health support, therapy or dependable medical guidance. Treat generated statements as material to check, particularly before sharing them with others.
A simple example
Fictional illustration, not a test: An adult hobbyist wants an assistant to brainstorm names for an imaginary board-game club. The setup could request five friendly names, each under four words, with no references to existing franchises.
Start with invented details: the fictional club likes cooperative games and meets on a made-up island. Ask for a short explanation of each suggestion, then request a revision that removes puns. Review whether the output respects the word limit and whether the explanations actually match the names. If evaluating voice, use the same harmless text and check intelligibility in your own environment. No real member information is needed.
The exercise gives you a small, repeatable evaluation without making the bot responsible for customer commitments or important decisions. A successful brainstorming exchange would still be limited evidence about broader reliability.
Who should consider it?
For a personal project, start small and decide whether the experience justifies its cost. For customer-facing work, make pricing clarity, data handling and a human fallback part of the selection process. Resolve those questions before uploading business information or presenting an agent as a dependable representative of your organization.
For another multi-bot option, read our Poe overview and compare the budget you would need for the same small set of tasks.
