FAQ

Can we use our own OpenAI or Anthropic key?

Yes. Customer-managed AI mode is available when finance or security needs direct provider visibility.

• Provider ownership
• Usage visibility
• Security review

Does Dchat replace Zendesk or Intercom?

Not for every team. Dchat is strongest as a website-first AI support layer with governed handoff, not as a full-suite replacement.

• Website-first
• Handoff
• Configured intake

Can AI perform account actions?

Dchat can use governed tools for context and drafted work, but sensitive changes should require human approval.

• Read context
• Draft work
• Approve changes

Do you publish compliance certifications?

Only signed evidence should be treated as certification proof. Current trust material separates controls from unclaimed certifications.

• Controls
• Gaps
• Signed proof

Frequently asked questions

How is Dchat different from a typical hosted chat product?

Dchat keeps AI replies, same-thread human handoff, branding controls, knowledge tools, and the team inbox together in one workspace. It also lets teams choose between a Dchat-managed OpenAI key and a customer-managed key, which is still uncommon among mainstream support products.

How does the widget embed work?

One script tag with your widget token. Paste it into your site's layout once and every future change — branding, AI settings, routing — happens in the dashboard. No plugins and no repeated code edits.

What happens when AI cannot handle the question?

The conversation moves to the human queue with the full transcript attached, so the visitor never restarts from scratch. Handoff can be triggered by a visitor request, a routing rule, a handoff keyword, or an AI limit you define.

Can I use my own OpenAI or Anthropic key?

Yes. Dchat supports both a Dchat-managed key and a customer-managed OpenAI key. Per-workspace settings include model selection, system prompt, escalation rules, and grounding — switchable without rebuilding the widget.

How does the knowledge base work?

Three ways in: write articles directly, paste URLs to import, or run the site crawler. The AI answers from that managed knowledge set, which lives in the same workspace as your routing and inbox settings.

What can I customize in the widget?

Launcher shape, colors, welcome message, starter prompts, proactive behavior, and white-label settings are all configured in the dashboard.

How much does Dchat cost?

Free forever for one site with 500 AI messages per month. Paid pricing is per agent seat: $29/month per seat on Dchat-managed AI, or $15/month per seat if you bring your own OpenAI or Anthropic key, or run without AI. No usage credits and no per-resolution fees.

How does Dchat compare with Intercom, Zendesk, Freshchat, Crisp, and Tidio?

Those are all credible products. Dchat is best aligned when the job is website-first AI support with branded chat, same-thread handoff, AI key ownership choice, and a self-hosted path. Broader suites win when omnichannel, enterprise operations, ecommerce depth, or bigger app ecosystems matter more.

Do I need to replace my current help desk to try Dchat?

No. The cleanest evaluation is to leave email, phone, or social support where it already lives and launch Dchat on the website first. That lets the team test AI coverage, transcript quality, and human handoff without a full migration.

Is there a self-hosted option?

Yes. Dchat itself is cloud-hosted, but Zchat runs the same platform on your own infrastructure — useful for regulated industries, regional hosting requirements, or air-gapped networks.

What does the support team get?

A real-time inbox with typing indicators, canned responses with shortcuts, contacts, routing rules, proactive triggers, reports, offline message capture, knowledge base management, and workspace settings in one admin surface.

What should I validate before using Dchat in production?

Test your AI prompts, routing rules, and retention settings against real traffic during the free tier. The current release does not claim formal HIPAA or SOC 2 certification by default, so validate compliance requirements specific to your environment first.

What should I measure in the first 30 days after launch?

AI-only resolution rate, visitor satisfaction on AI-only conversations, time from handoff to first human reply, and how many weak answers turned into knowledge or routing fixes. Those four numbers tell you faster than raw volume whether AI-first support is holding up.