AI-native business operating platform
Nobody manually operates the business logic. Not the customer. Not your staff either.
One codebase, sold three ways — managed service, self-serve SaaS, and white-label reseller. Every business action runs through a typed, permissioned Automation Core, executed by a dedicated AI assistant per tenant.
One product, three revenue paths
Managed Service
We run it for them
Your ops team onboards the client through the internal console. The client's dedicated AI assistant runs day-to-day operations on their behalf — closest to done-for-you.
Self-Serve SaaS
They run it themselves
A business signs up on the public site, completes an AI-guided onboarding wizard, and operates its own instance through its assistant and dashboard — no human on your side involved.
White-Label Reseller
Partners run it under their brand
A partner gets their own branded instance — landing page, signup flow, and admin console — to onboard and manage their own customers, with revenue-share billing back to you.
Same core product throughout — the difference is who configures what, whose brand appears where, and how billing flows. Not three codebases.
The Automation Core
Every business action — scoring a lead, drafting a post, sending an invoice, updating a pipeline stage — is a typed, permissioned tool. The AI assistant and the dashboard UI are two clients of the exact same contract; there is no feature the AI can't touch that the UI can, or vice versa. Risk-tiering and audit logging happen once, centrally, not per feature.
Always autonomous
No approval gate, ever — and not tenant-configurable. Reading data or drafting a message carries no irreversible effect.
Autonomous by default
Updating a pipeline stage, tagging a lead. A tenant can tighten this to approval-required or denied from their own settings console.
Approval-required by default
Sending an invoice, an external email, a payment. A tenant can explicitly opt into full autonomy here — their informed choice, not ours.
Every action is attributed, hash-chained, and immutable at the database level — the audit log cannot be edited or deleted, from any role, by design.
Agnostic where it matters
Provider-agnostic
Every AI feature goes through one internal AiProvider interface with adapters for Anthropic, OpenAI, Google, and self-hosted inference. No feature code imports a vendor SDK directly. Swap the model, keep the product.
Industry-agnostic
No hardcoded "if industry === X" branches. Pipeline stages, website sections, tone, and fields come from a data-driven Industry Profile the AI can synthesize for a business type it has never seen before.
Region-agnostic
Currency, tax rules, language, and timezone are per-tenant configuration resolved at onboarding. Compliance rule packs (e.g. a country's e-invoicing format) are pluggable modules, never baked into core logic.
Isolated by tenant, not by convention
Platform → Reseller → Tenant → Tenant Team Member, plus a first-class AI Assistant actor identity per tenant. No tenant's data, conversation history, or credentials are reachable from another tenant's context under any code path — enforced at the data-access layer and covered by tests, not just convention.