> ## Documentation Index
> Fetch the complete documentation index at: https://docs.prisme.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Platform Overview

> Where you are in your Gen.AI journey, and why the foundation has to be a neutral, resilient layer no single cloud or model vendor controls.

The [introduction](./introduction) covered *why* Prisme.ai is a unified foundation rather than another silo. This page is the **strategic view**: where you are in your Gen.AI journey, and why that foundation has to be a **neutral, resilient layer** that no single cloud or model vendor controls.

## Where are you in building your AI Information System?

Most enterprises are past experimentation. The work now is **scaling to measurable ROI**, with governance and freedom of choice. The platform meets you at your stage and grows with you, from secure chat to a scalable, AI-by-design Information System.

<Frame>
  <img src="https://mintcdn.com/prismeai/h62RQD_vQ0CV0civ/images/maturity-model.svg?fit=max&auto=format&n=h62RQD_vQ0CV0civ&q=85&s=53dd2f80888d16e9e6209acdf12953a7" alt="Building your AI Information System on a value-at-scale axis. 1 Adopt: secure multi-LLM chat for everyone (Chat, Store). 2 Specialize: agents that act, with your skills and knowledge (Agent Creator, tools and MCP, Knowledges). 3 Scale and transform: custom apps, mutualized infrastructure, lower software costs (Builder, Collections, connectors and MCP). At every stage: governance central and delegated, sovereignty as freedom of cloud and model, and resilience." width="1200" height="560" data-path="images/maturity-model.svg" />
</Frame>

* **Adopt**: give everyone secure, multi-LLM chat. Individual productivity, no data leaving your control.
* **Specialize**: build agents that **act**, connected to your **skills** (tools, MCP) and your **knowledge**, multi-LLM and cloud-resilient. Moving from chat to agents that get things done.
* **Scale & transform**: build custom apps and workflows with [Builder](/products/ai-builder/overview), mutualize infrastructure securely, and **rationalize internal software costs** by replacing siloed or underused applications, with measurable ROI.

## Resilience and freedom of choice

A foundation earns its place by being the one layer no single vendor controls. This is a real **architecture** property, not a slogan: the platform is event-driven and layered to stay cloud- and model-agnostic. The low-level architecture, services, and databases are covered in [Self-Hosting](/self-hosting/overview).

<CardGroup cols={2}>
  <Card title="LLM resilience" icon="brain">
    Models change fast: prices move, quality shifts, and some are deprecated (GPT-4o was retired). Prisme.ai is natively multi-model and routes across OpenAI, Anthropic, Mistral, AWS Bedrock, Google Vertex, and self-hosted models. Swap or route per use case **without touching your agents**.
  </Card>

  <Card title="Cloud resilience" icon="cloud">
    Run on any major cloud or on-premises: AWS, Azure, Google Cloud, OVHcloud, Outscale, OpenShift. No cloud-specific service locks your agents, data, or governance in.
  </Card>
</CardGroup>

Here, **sovereignty means resilience and freedom of choice, not protectionism**: you stay free to pick the best cloud and the best model, and to switch, without a rewrite.

### Why a cloud provider or an LLM provider cannot be your AI foundation

* **A cloud provider's AI stack ties you to that cloud.** Your agents, data, and governance become cloud-specific, and leaving means rebuilding. Their incentive is to keep you on their cloud.
* **An LLM provider's platform ties you to that model.** When a model is deprecated or repriced, you are exposed, and you cannot route to a competitor. Their incentive is to keep you on their model.
* **Neither is neutral.** Their business is selling their cloud or their model. A foundation's job is the opposite: to keep you free to choose per use case, and to switch **without a rewrite**.
* **Only a vendor-neutral layer can govern across clouds and models.** Governance, traceability, security, and cost control have to sit *above* any single provider to be real. That neutrality is the product.

## Governance that scales with you

Governance is itself a maturity dimension. It starts **central** and becomes **delegated** as you scale, so teams get the freedom to act within guardrails rather than waiting on a bottleneck. That balance (central rules, per-organization autonomy) is what lets adoption grow without losing control.

<CardGroup cols={2}>
  <Card title="Governe" icon="shield-check" href="/products/ai-governance/overview">
    Central and per-org governance: SSO, roles, model access, audit, quotas, and retention.
  </Card>

  <Card title="Adoption & Governance playbook" icon="users-gear" href="/resources/adoption-governance/overview">
    Who does what, one org or several, and how to open the platform without chaos.
  </Card>
</CardGroup>

## Build and deploy

The foundation carries a full engineering lifecycle. Go straight to the detail:

<CardGroup cols={3}>
  <Card title="Build" icon="screwdriver-wrench" href="/dsul/overview">
    Workspaces as React front-ends and YAML automations, from VS Code over MCP, versioned in Git.
  </Card>

  <Card title="Deploy" icon="rocket" href="/resources/adoption-governance/deployment">
    Platform environments with CI/CD, and per-use-case dev and prod workspaces synchronized by Git.
  </Card>

  <Card title="Architecture" icon="sitemap" href="/self-hosting/overview">
    Event-driven services, layering, and databases for teams running the platform.
  </Card>
</CardGroup>

## Next steps

<CardGroup cols={3}>
  <Card title="Products & Capabilities" icon="puzzle-piece" href="../products/overview">
    The building blocks of the platform
  </Card>

  <Card title="First Steps Guide" icon="person-walking" href="/get-started/first-steps">
    Getting hands-on with the platform
  </Card>

  <Card title="Account Setup" icon="user-gear" href="/get-started/account-setup">
    Configure your account and environment
  </Card>
</CardGroup>