Skip to content
Ektasi.
Platform / Self-hosted AI & Data Sovereignty

Your AI runs where your data lives.

Most “AI platforms” are a thin wrapper that ships your prompts and your customers' data to someone else's servers. Ektasi's AI suite can run entirely on infrastructure you own, in India — so intelligence never costs you sovereignty.

Why it matters now

DPDP changes the default answer to “where does the data go?”

India's Digital Personal Data Protection Act moves enforcement forward, and the penalties for mishandling personal data are steep. When a regulator asks where your customers' data — and the AI processing it — physically sits, “a US cloud AI vendor” is an expensive answer.

Ektasi's architecture makes the compliant path the default: keep the data in India, keep the AI on your own server, and keep an immutable record of every access. You are not bolting compliance on afterwards — it is how the platform is built.

Three ways to run the AI

Fully self-hosted

Point the platform at an OpenAI-compatible endpoint you run yourself — Ollama, vLLM, or any compatible gateway on your own hardware.

Prompts and data never leave your network.

Bring your own key (BYOK)

Use your own OpenAI, Anthropic, Google Gemini, or OpenRouter keys, encrypted per tenant, with automatic failover between providers.

You control the vendor relationship and the spend.

Hybrid routing

Route cheap text tasks to a platform model and sensitive workloads to your self-hosted endpoint — per feature, per tenant.

Cost where it is fine, sovereignty where it matters.
System map · Self-Hosted AI
Self-Hosted AIways to runFully self-hostedBring your own key(BYOK)Hybrid routing

The platform speaks an OpenAI-compatible protocol, so new providers and self-hosted engines slot in without changing your workflows.

What sovereignty actually means here

Data residency in India

Deploy on infrastructure located in India so personal data stays within the country by design.

Per-tenant AES vault

Provider keys and secrets are encrypted with AES-256-GCM using per-tenant, HKDF-derived keys — never stored in plaintext.

Row-level tenant isolation

PostgreSQL row-level security means a query scoped to one tenant cannot return another tenant’s data.

Append-only audit ledger

Every sensitive action, including AI usage, is written to a database-enforced, append-only audit trail.

No forced data sharing

Self-hosted models mean prompts, documents, and outputs are not sent to a third-party AI vendor unless you choose BYOK.

Cryptographic erasure

Decommissioning a tenant destroys its per-tenant salt, rendering its ciphertext — including backups — permanently unreadable.

The outcome

A full AI-native business platform your compliance team can actually sign off on — because the data, and the model, stay under your control.

Keep the intelligence. Keep the sovereignty.