Frequently asked questions
Can't find what you're looking for? Send us a message and we'll get back to you.
Privacy & Data
Does Canary use my customers' conversations to train AI models?
No. Conversation data is never used to train, fine-tune, or otherwise improve any AI model — ours or anyone else's. Your customers' support conversations belong to you, and that is the only place they are used.
When the assistant answers a question, it processes the conversation in real time and returns a response. The AI provider we route through operates under a zero-retention agreement: they do not store or log the content of those calls after the response is delivered.
Where is my conversation data stored?
Conversation data is stored in infrastructure hosted in the United States. Data is encrypted in transit (TLS) and at rest. Workspace owners can export or permanently delete all conversation history at any time from their dashboard — no support ticket required.
Do you share my data with third parties?
We do not sell, rent, or share your data with third parties for their own marketing or training purposes. We work with service providers (cloud hosting, email delivery, analytics) who access data only to perform their services and are contractually obligated to protect it. For a full list of categories, see our Privacy Policy.
Is Canary GDPR and CCPA compliant?
We are committed to compliance with applicable data protection laws including GDPR and CCPA. Workspace owners can export or delete end-user conversation data on demand, and we support data subject access requests. If you need a Data Processing Agreement (DPA) for your deployment, please reach out to us at privacy@assistantcanary.com.
Can I delete all conversation history?
Yes. Workspace owners can permanently delete all conversation history for any client from the dashboard. You can also set a rolling retention window so history older than a chosen number of days is automatically purged.
How the AI Works
How does the assistant decide what to say?
Every answer is grounded in the knowledge base you build. When a customer asks a question, the assistant searches your uploaded content — FAQs, product docs, support articles, brand guidelines — and constructs a reply from those sources. If the answer is not in your knowledge base, the assistant says so rather than guessing. It cannot and does not draw on information from other companies' customers.
What stops the assistant from making things up?
Our guiding principle is: source-cited or silent. The assistant only states something if it can point to a passage in your knowledge base that supports it. When it cannot find a solid source, it tells the customer it does not know and offers to escalate to a human agent. Operators can also configure topic guardrails that block specific subjects outright.
Can the assistant match my brand voice?
Yes — this is a core feature, not an afterthought. During onboarding you define tone settings, a greeting style, and example responses. You can also upload brand guidelines directly to the knowledge base. The assistant is instructed to write in your style, not in generic AI prose.
What happens when the assistant cannot help?
You choose the escalation path. Options include collecting the customer's details for a callback, routing to a live-chat queue, or directing them to a support email. The assistant clearly signals to the customer when it is handing off, so there is no awkward dead-end.
Does Canary support languages other than English?
Yes. The assistant can understand and respond in the language your customer writes in, as long as your knowledge base contains content the assistant can reference. If your knowledge base is English-only, answers will be translated but accuracy depends on the quality of the source material.
Plans & Billing
Is there a free trial?
Yes. All new workspaces start on a trial that includes a generous message allowance so you can fully test the assistant with real traffic before committing to a paid plan. No credit card is required to start.
How is pricing calculated?
Pricing is based on the number of AI-handled conversations per month. You pick a plan tier that matches your expected volume; if you go over, we automatically scale rather than cut off your customers. See the Pricing page for current plan details.
Can I cancel at any time?
Yes. There are no long-term contracts. You can downgrade or cancel your subscription at any time from the billing section of your workspace. Your data remains accessible until the end of the billing period, after which it is deleted per our retention policy.
Can I run the assistant for multiple brands or clients?
Yes. Each workspace can manage multiple clients, each with its own knowledge base, branding, and configuration. This is designed for agencies and operators who support several brands from a single account. Each client's data is fully isolated from the others.
Getting Started
How long does it take to go live?
Most teams have a working assistant within a day. The fastest path is to paste in a few dozen FAQ answers, configure the greeting and tone, then embed the widget. You can keep adding and refining content in the background while it handles real conversations.
How does the widget get added to my site?
You copy a single script tag from the Command Center and paste it into your site's HTML, or install our npm package if you prefer. No server-side changes are required. The widget loads asynchronously and does not affect your page's Core Web Vitals.
What support is available during setup?
All plans include access to our documentation and onboarding guides. Paid plans also include email support with a guaranteed response time. If you want a hands-on onboarding call, get in touch — we are happy to walk through your specific setup.