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

# Senderz AI

> What the AI features are, where each one lives in the product, and what they cost you.

**Senderz AI** is the second item in the sidebar. It opens the AI assistant, a chat that can read your workspace and act in it.

AI is not only that one screen. Smaller AI helpers are built into the screens where you already work: the segment builder, the email and SMS editors, the popup builder, the coupon library, the brand kit and the contact profile. They all run on the same provider key.

<Note>
  Every AI feature runs on **your own provider API key**. Senderz does not resell
  AI tokens and there is no AI plan to buy. Your provider bills you directly for
  what you use.
</Note>

## Before anything works

You need one AI provider connected. Until then, every AI control in the product shows a short "Connect an AI provider in Settings" message instead of running.

<Card title="Connect your AI provider" icon="key" href="/en/integrations/ai-provider">
  Choose Claude, Gemini, OpenAI or a compatible endpoint, paste your key, test
  it and pick a model. Owners and admins only.
</Card>

Once a key is saved, everyone on the team can use AI features. Only owners and admins can change the key.

## Where each AI feature lives

<AccordionGroup>
  <Accordion title="AI assistant" icon="comments">
    **Sidebar → Senderz AI.** A chat that reads your contacts, campaigns,
    automations, segments, analytics, deliverability, wallet and store data, and
    can create and change things for you. It asks before anything destructive or
    costly.

    You can also reach it from the **Marketing Agent** button in the top bar,
    from the command palette ("Ask the Marketing Agent"), from the **Ask AI**
    card on a contact profile, and from the **Ask AI about your analytics** bar
    on the Analytics page. Those last two open the chat with your question
    already filled in.

    [Read the full assistant page](/en/ai/assistant)
  </Accordion>

  <Accordion title="Build a segment from a description" icon="filter">
    **Segments page**, and inside the campaign wizard audience step and the
    automation segment picker. Describe the audience in plain language and get a
    filter you can review before saving.

    The builder checks your real tags, consent counts and predictive tiers
    before it proposes anything, and tells you in the caveats when part of your
    request cannot be expressed as a filter.

    See [Segments](/en/audience/segments).
  </Accordion>

  <Accordion title="Write and rewrite content" icon="pen-nib">
    * **Generate with AI** in the email builder writes a full email from a
      one-line brief, matched to your brand kit.
    * **Write with AI** does the same for an SMS body.
    * **Suggest with AI** rewrites the text you have selected. It appears on the
      campaign wizard, the automation message panel, contact notes, popup copy
      and inside the rich text toolbar.

    See [Templates](/en/messaging/templates).
  </Accordion>

  <Accordion title="Generate images" icon="image">
    Any image field that opens the image picker has a **Generate** tab. Describe
    the image, choose square, portrait or landscape, and the result is saved
    into your media library.

    Only OpenAI and Gemini can generate images. On a Claude connection the tab
    tells you so instead of failing.
  </Accordion>

  <Accordion title="Brand voice and business summary" icon="palette">
    **Media → Brand.** **Generate with AI** turns a sample of your own writing
    into voice rules, and drafts a short summary of your business. Both are then
    used to ground later copy generation, so generated email matches how you
    actually write.
  </Accordion>

  <Accordion title="Draft a coupon" icon="tag">
    **Products → Coupons.** The **Draft with AI** bar takes a sentence such as
    "15% off orders over 200 shekels until the end of the month" and pre-fills
    the create form. Nothing is saved until you confirm.

    See [Coupons](/en/messaging/coupons).
  </Accordion>

  <Accordion title="Popup copy suggestions" icon="window-restore">
    **Popup builder.** Heading, subheading and button fields carry a suggest
    control that offers alternative wording for that one field.

    See [Popups](/en/messaging/popups).
  </Accordion>

  <Accordion title="A/B test suggestions" icon="flask">
    **A/B tests page.** Suggests tests worth running based on what it can see in
    your workspace.
  </Accordion>

  <Accordion title="Pre-send checks" icon="clipboard-check">
    Ask the assistant to audit a campaign, an automation, a template or an SMS
    before you send. The checks themselves are deterministic, not guesses: a
    missing template, an empty audience, an unverified sending domain, a
    schedule in the past, a missing subject line, images with no alt text,
    broken or insecure links, SMS credit cost, an automation with no trigger or
    no message step.

    These checks currently run through the assistant chat. There is no separate
    audit button on the campaign, template or automation screens yet.
  </Accordion>
</AccordionGroup>

## The AI planner

There is a marketing calendar planner at the address `/ai/planner`. You describe what you want to promote and a date range, and it drafts a calendar of campaigns you can accept one by one into normal drafts.

<Warning>
  **The planner has no menu entry today.** It is not in the sidebar, the command
  palette or any button, so the only way to open it is to type the address. Treat
  it as unfinished until it appears in navigation.
</Warning>

## What AI costs you

Senderz adds nothing to the price. You pay your provider for the tokens each request uses.

* **Usage this month** in the AI configuration panel shows estimated cost, total tokens and a per-model breakdown for the current calendar month.
* That figure is calculated from published model rates. It is an estimate, not an invoice. Your provider's own billing dashboard is the authoritative number.
* If you used a model with no published rate, the card says so, and your real total is higher than shown.

Two things drive cost more than anything else: the model you pick, and the **Reasoning** setting. Both are in [AI provider settings](/en/integrations/ai-provider).

<Warning>
  **Budget alerts are not available yet.** Senderz can compare estimated spend
  against a monthly budget and notify you at 80% and 100%, but no screen lets
  you set that budget, so the alerts never fire. Watch your provider's billing
  dashboard instead.
</Warning>

## What AI is never allowed to do

Every AI action runs through the same code as the equivalent button in the interface. That means the rules you rely on still apply, whether a person or the assistant started the action.

| Rule                            | Still enforced                                                           |
| ------------------------------- | ------------------------------------------------------------------------ |
| Recipient consent               | Yes. AI cannot mail someone who has not opted in.                        |
| Suppression list                | Yes. Unsubscribes, bounces and STOP replies still block the send.        |
| Israeli quiet hours and Shabbat | Yes, at send time, whatever schedule was set.                            |
| Verified sending domain         | Yes. An unverified From address is still rejected.                       |
| SMS credit wallet               | Yes. Every SMS is still paid for from your balance.                      |
| Campaign approval policy        | Yes. If approvals are on, an AI-scheduled campaign still needs sign-off. |
| Workspace isolation             | Yes. The assistant can only ever see your own workspace.                 |

AI never sends a campaign on its own. It creates drafts and, when you ask it to, schedules one. The send pipeline decides what actually goes out.

See [Consent](/en/audience/consent), [Quiet hours and Shabbat](/en/compliance/quiet-hours) and [Israeli compliance](/en/compliance/israel).

## When AI stops working

If your provider rejects a request, Senderz stops. There is no shared fallback key, by design.

You get a notification in the bell naming the cause: key rejected, no key on file, key lacks access, rate limited, out of quota, or endpoint unreachable. The connection badge in Settings shows the same state with your provider's own response quoted.

<Note>
  These are in-app notifications only. Senderz does not email you when a key
  fails, so check the bell if AI has gone quiet. Everything else in Senderz keeps
  working normally: campaigns, automations and sending are unaffected.
</Note>

## Related

<CardGroup cols={2}>
  <Card title="AI assistant" icon="comments" href="/en/ai/assistant">
    What the chat can read, what it can change, and what it asks before doing.
  </Card>

  <Card title="AI provider settings" icon="key" href="/en/integrations/ai-provider">
    Keys, models, reasoning, usage and disconnecting.
  </Card>

  <Card title="MCP access" icon="robot" href="/en/integrations/mcp">
    Let an outside AI client work in your workspace over MCP.
  </Card>

  <Card title="Segments" icon="filter" href="/en/audience/segments">
    Where the plain-language segment builder lives.
  </Card>
</CardGroup>