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

# AI Models & Bring-Your-Own-Key

> Choose how AI inference is powered for your organization — use included credits, or bring your own OpenAI or Anthropic API key.

Founder Sherpa runs every AI feature — transcript parsing, evidence analysis,
hypothesis quality checks, the CoPilot assistant, and more — through its own
backend, over your organization's data. You control how that inference is
**powered and billed** with a per-organization **inference mode**.

## Inference modes

<CardGroup cols={2}>
  <Card title="Hosted (default)" icon="cloud">
    AI runs on Founder Sherpa's platform API keys and is metered against your
    organization's included AI credits. Nothing to configure — it works out of
    the box.
  </Card>

  <Card title="Bring your own key (BYO)" icon="key">
    AI runs on **your own** OpenAI or Anthropic API key. You pay your provider
    directly for token usage, and AI credits no longer apply to your
    organization. Best for cost-conscious power users who already have a
    provider account.
  </Card>
</CardGroup>

In both modes, **inference still runs through Founder Sherpa's backend over your
data and is rendered in the app** — bring-your-own-key only changes who pays the
provider for tokens. Your evidence graph, methodology, and synthesis stay in
Founder Sherpa.

## Setting your inference mode

<Note>
  The **AI & Models** settings tab is available to organization **owners and
  admins**. Members don't see it.
</Note>

<Steps>
  <Step title="Open AI & Models settings">
    From the sidebar, go to **Organization Settings**, then open the **AI &
    Models** tab.
  </Step>

  <Step title="Choose a mode">
    Toggle between **Hosted (credits)** and **Bring your own key**.
  </Step>

  <Step title="Add a provider key (BYO only)">
    When BYO is selected, enter an API key for the provider you want to use —
    **OpenAI** and/or **Anthropic**. Click **Save**. The key is validated and
    stored encrypted; you can **Remove** it at any time.
  </Step>
</Steps>

## Supported providers

| Provider  | Hosted                           | Bring-your-own-key |
| --------- | -------------------------------- | ------------------ |
| OpenAI    | Yes (default)                    | Yes                |
| Anthropic | Yes (if enabled by the platform) | Yes                |

<Note>
  Support for generic OpenAI-compatible endpoints (custom base URLs, e.g. Azure
  OpenAI or self-hosted gateways) is on the roadmap.
</Note>

## How your keys are protected

<CardGroup cols={2}>
  <Card title="Encrypted at rest" icon="lock">
    BYO API keys are encrypted with AES-256-GCM before they're stored. The
    encryption key lives only in our server environment — never in the database.
  </Card>

  <Card title="Never shown to the browser" icon="eye-slash">
    A saved key is never sent back to your browser or displayed again. The
    settings screen only shows whether a key is set and its last four characters.
  </Card>

  <Card title="Server-side only" icon="server">
    Keys are decrypted only inside Founder Sherpa's backend at the moment an AI
    request is made. They're held under the same strict, service-only access as
    our most sensitive data.
  </Card>

  <Card title="Owner / admin controlled" icon="user-shield">
    Only organization owners and admins can set, replace, or remove a key.
  </Card>
</CardGroup>

## What changes in BYO mode

* **Billing** — token usage is billed by your provider, not against Founder
  Sherpa AI credits. The credits panel and balance indicator are hidden for the
  organization while BYO is active.
* **Usage visibility** — your month-to-date AI usage (tokens and call counts) is
  still shown in the app for observability; it's just not charged to credits.
* **If a key is missing** — if BYO mode is on but no key is configured for the
  provider an AI feature needs, that action stops with a clear message asking you
  to add a key in **Settings → AI & Models**. It will never silently fall back to
  platform billing.

## Switching back to hosted

Switch the mode back to **Hosted (credits)** at any time. AI immediately resumes
running on platform keys and metering against your credits. Your saved BYO keys
remain stored (encrypted) unless you remove them, so you can switch back to BYO
later without re-entering them.

<Note>
  See [Security & Data Protection](/docs/security) for how interview content and
  credentials are handled, and [AI Features](/docs/features/ai-features) for what
  each AI capability does.
</Note>