Who controls the provider relationship?
Dchat can run with a Dchat-managed key for speed or a customer-managed OpenAI key for governance. Teams can switch modes without replacing the widget, inbox, or reporting flow.
Trust the product by
seeing the boundaries clearly.
Dchat's trust story is not "we do everything." It is clearer than that: visible AI ownership, same-thread human fallback, approval-safe actions, and an open deployment path if cloud-only stops fitting later.
Bring your own OpenAI key
Self-hosted path through Zchat
Approval-safe outbound actions
Clear live-now vs next scope
Control surfaces
Control what matters before you scale traffic.
Procurement and operations questions usually collapse into a small set of concerns: who owns the model relationship, what happens when AI is unsure, where data can live, and which actions remain human-reviewed.
AI ownership choice
Human review path
Cloud or self-hosted
What buyers need answered
The trust questions that actually decide the shortlist.
These are the questions Dchat can answer clearly today without pretending to have broader-suite depth it does not yet ship.
What happens when AI should stop?
Dchat keeps the fallback explicit: visitors stay in the same thread, agents inherit the transcript, and outbound actions can stay queued for review instead of firing automatically.
Can the operating model change later?
Yes. Cloud is the default path, but the product family keeps a self-hosted route through Zchat when residency, procurement, or private-network requirements appear later.
What is not being claimed?
Dchat should be evaluated as website-first AI support with a light action layer. It is not yet the right answer for teams that need full omnichannel coverage, deep mobile workforce tooling, or enterprise compliance packaging on day one.
Current posture
What Dchat can say clearly today.
This is the product-trust split that matters during rollout and procurement review.
| Area | Current Dchat position | What to validate yourself |
|---|---|---|
| AI provider control | Choose Dchat-managed AI or a customer-managed OpenAI key. | Model selection, prompt behavior, rate limits, and provider-side billing policies. |
| Human escalation | Same-thread takeover with transcript continuity and task review flows. | Handoff rules, staffing expectations, and response-time policies. |
| Outbound actions | Approval-safe webhook actions and notification endpoints are live. | Which actions should remain manual, reviewed, or environment-specific. |
| Deployment path | Dchat cloud now, Zchat self-hosted when infrastructure control matters. | Jurisdiction, residency, internal-network, and procurement requirements. |
| Compliance posture | No blanket claim of HIPAA, SOC 2, or full enterprise certification in the current release. | Industry-specific legal, security, retention, and audit requirements before production rollout. |
Live now vs next
Buyers should not have to infer roadmap from marketing copy.
A credible trust page makes the product boundary visible. That reduces bad-fit deals and makes good-fit evaluations move faster.
- Website widget with AI replies
- Same-thread human handoff
- Knowledge import, crawl, and hosted help content
- Slack, Teams, Discord, and webhook notifications
- Approval-safe outbound action queue
- AI key ownership choice
- Email inbox support
- CRM and account context inside support workflows
- Order, billing, and commerce lookups
- Broader async channel packaging after email
- Omnichannel support on day one
- Large marketplace and mobile agent app depth
- Formal enterprise compliance packaging up front
- Cross-department no-code AI automation as the core product
Recommended validation
What a careful team should test before launch.
These checks matter more than a generic "enterprise ready" label.
Test prompt quality, KB grounding, and handoff timing against real support questions.
Review retention, data handling, and provider configuration against internal policy.
Decide which outbound actions can stay auto-notified and which must require human approval.
Confirm whether cloud fits now or whether self-hosted review should happen before rollout.
Next step
Evaluate the product with the hard questions up front.
If Dchat fits, it should fit for clear reasons: website-first support, visible AI control, same-thread human fallback, and a deployment path that stays open later.