Straight answers.
No AI theatre.

No. The product is being built as an AI Business Operator. The AI Receptionist is the human-facing front door; specialist agents, business intelligence, tools, memory and governed execution operate behind it.

On the public homepage you can ask Michelle about Centrix One, the plans, how voice works and which page answers your question. Inside the workspace, the long-term interface is ordinary business language: ask about customers, sales, marketing, operations, documents, costs, appointments, locations or business performance. Centrix prepares work and uses approval controls for material external actions.

The strategy is to reduce fragmentation, not force a business to replace everything on day one. Centrix is being built to connect to approved systems and progressively become the operating layer across them.

Yes, that is a core Business Analyst use case. The planned workflow extracts evidence from bills and statements, calculates comparable metrics, detects anomalies and produces evidence-backed savings opportunities.

Yes. Centrix can model authorised premises data such as area, rent, rates and utilities, then calculate unit economics and compare locations or periods where the underlying data supports the comparison.

Yes. Recommendations will be tied to evidence and clearly labelled as recommendations or estimates. Centrix should never present an estimate as a verified saving.

Centrix is provider-agnostic. The model router is intended to select among approved providers such as Google Gemini, OpenAI, DeepSeek, Kimi, Claude, Grok and Qwen according to capability, quality, latency, cost, privacy and tenant policy.

Yes, the architecture includes place search, place details, geocoding, address validation, routes and dashboard map cards. The required Google services will be connected progressively.

Yes. HighLevel exposes OAuth 2.0 and APIs covering contacts, conversations, opportunities, calendars, payments, invoices, social publishing and other capabilities. Centrix will treat HighLevel as an integration, not as the product identity.

The integration architecture is designed for it. Social connections should use platform/HighLevel OAuth and tenant-scoped permissions, with publishing treated as a governed external action.

For low-risk read-only work, tenant policy can eventually allow automatic execution. Material external actions, financial commitments, messages, purchases and destructive actions remain subject to the configured governance and approval policy.

The proposed public offer includes a 7-day trial with controlled usage. Trial limits are designed to let a business experience the product without exposing Centrix to uncontrolled AI, voice or third-party costs.

The recommended launch model is a no-card trial. This reduces friction while keeping usage caps in place.

No. Centrix will separate included product usage from metered provider costs where applicable and show usage or additional charges clearly before paid execution is enabled.

Yes, public pricing can evolve as capabilities and operating costs mature. Founding customers should receive a clearly documented price-protection commitment rather than an ambiguous promise.