Straight answers, grouped by topic.

No. Neocortex is built on standard Next.js and Postgres (Supabase). Your data exports in open formats, the code runs on any major cloud, and with escrow or a source licence you hold the code itself.

Yes, both. License the source outright for your own deployment, or subscribe with the source held in independent escrow.

You do. Every deployment is a dedicated instance — never a shared multi-tenant pool — and your data is never used to train anyone else’s model.

Either. Use it as SaaS we run, as a platform you extend with the App SDK, or as software you run yourself — same product in every model.

Yes. Application, database, files and backups can all stay in Australian regions.

Yes. Application, database, files, backups and AI inference can be pinned to any region you choose for full data residency and sovereignty.

Yes. It ships as a standard container for your data centre, your own Azure, AWS or Google Cloud tenancy, or a private network — with local models if no prompt may leave your walls.

AWS, Azure, Google Cloud and Vercel, or your own infrastructure. The stack is Next.js with a Supabase (Postgres) database, shipped as a standard container.

Neocortex suits organisations of any size. It scales horizontally and can be geo-replicated across regions, with failover in seconds.

The platform is architected against the control themes of ISO/IEC 27001 and the five pillars of the Well-Architected Framework: access control, secrets management, audit logging, least privilege, and encryption in transit and at rest. Formal certification is available as a customer-funded engagement; the architecture does not need to change to get there.

Yes. Customers are welcome to commission an independent penetration test against their deployment, and we will support it.

Neocortex is a single-operator application today. Authentication is one shared operator PIN by design. It is not presented as per-user accounts, MFA, SSO or a hardened multi-tenant SaaS identity system. The organisations data model exists as a foundation for future multi-organisation capability, but it is not the current authentication boundary.

Any of them: OpenAI, Anthropic Claude, Google Gemini, Azure OpenAI, Mistral, DeepSeek, xAI, Groq, OpenRouter and models you host locally. Choose per task, set budgets, and fail over automatically.

Neocortex is designed around explicit operator controls: AI usage can be monitored against budgets, alerts can surface approaching limits, hard limits can stop further consumption, and the router can optimize model selection and token use. Caching can avoid repeating eligible work. These controls should be evaluated against the exact deployment configuration rather than treated as a universal guarantee.

Only where you configure it to. Route to the providers you approve, your own Azure OpenAI deployment or models you host yourself.

Yes. Name, logo, colours, domain, documents, email and even the AI persona are configurable for a deployment.

Implementation depends on the deployment, integrations and controls required. Neocortex is currently introduced through guided engagements rather than a self-serve SaaS signup.

Usually, yes. We map each legacy app to Neocortex modules, migrate the data, and rebuild what is unique on the App SDK — replatforming several apps within months rather than years.

For up to 500 users, pricing follows scope, users, hosting and support. Above 500 users, contact us for enterprise pricing.