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

# Configure Sally's Reply Handling Tone and Knowledge

> Set up tone sliders, a knowledge base, auto-send rules, and runtime policies so Sally handles inbound replies accurately and in your voice.

A reply handler is the playbook Sally follows when a prospect writes back. When a reply lands, she reads it, selects the best-matching handler profile, and drafts an answer in your voice — drawing only from the knowledge you have approved, and following the runtime rules you have set for scenarios like positive responses, out-of-office messages, meeting bookings, and unsubscribe requests.

Reply handlers live under **AI Setup → Reply Handlers** in the app. Each profile covers a complete set of instructions: tone, knowledge base, auto-send preferences, runtime policy, and hard-stop topics. You can maintain multiple profiles and keep each one in **Live** or **Draft** state, with one designated as the account default.

## What you can do

* **Run multiple profiles** — keep separate reply playbooks for different products, audiences, or campaigns, and flip each between **Live** and **Draft**.
* **Set one account default** — the profile Sally falls back to when no other profile is a better match for an incoming reply.
* **Shape tone with four sliders**: Warmth, Length, Formality, and Salesiness.
* **Connect a knowledge base** — docs sites, Notion spaces, or PDFs Sally is allowed to quote, with per-source re-indexing controls.
* **Add reference replies** — real prospect exchanges Sally studies to calibrate your voice.
* **Build an objection library and competitive advantages** so Sally has rebuttals and positioning ready when needed.
* **Upload collateral docs** whose extracted text feeds the profile's sales pitch.
* **Control auto-send** — choose which reply types Sally can send on her own versus draft for your review.
* **Tune the runtime policy** — six classifier handlers that decide what happens to the sequence when a particular reply type arrives.
* **Set "Never reply about" topics** — hard-stop phrases that route a thread to a review channel instead of generating a draft.
* **Track 30-day performance** — drafts written, auto-sent, edited by you, and reply-to-meeting conversion.

## How to create a reply handler profile

<Steps>
  <Step title="Open Reply Handlers">
    Go to **AI Setup → Reply Handlers**. Existing profiles appear as pills across the top, each showing a status dot and a **Live / Draft** toggle.
  </Step>

  <Step title="Create a new profile">
    Click **New profile**, or **+ Add your first profile** if you have none yet. The profile creation drawer opens.
  </Step>

  <Step title="Name and save the profile">
    Give the profile a clear name you will recognize in the switcher — for example, the product or audience it covers. Click **Save** to add it to the pill strip.
  </Step>

  <Step title="Configure tone, knowledge, and rules">
    Select the profile pill to open it, then work through each section: set the tone sliders, connect knowledge sources, configure auto-send preferences, and tune the runtime policy (each covered in detail below).
  </Step>

  <Step title="Set it Live when ready">
    New profiles start as **Draft**. When you are satisfied with the configuration, click the **Live / Draft** toggle on the pill to set it **Live** so Sally can use it on incoming replies.
  </Step>
</Steps>

<Tip>Use **Make default** to designate the active profile as the account default. Only one profile can be the default at a time — setting a new default automatically clears the flag from any previously designated one.</Tip>

## How to shape tone

<Steps>
  <Step title="Select the handler profile">
    Click its pill in the switcher. The four tone sliders appear at the top of the left column.
  </Step>

  <Step title="Adjust each slider">
    Drag each slider to shape Sally's voice across four axes:

    | Slider         | Range           |
    | -------------- | --------------- |
    | **Warmth**     | Crisp → Warm    |
    | **Length**     | Short → Long    |
    | **Formality**  | Casual → Formal |
    | **Salesiness** | Friend → Pitch  |
  </Step>

  <Step title="Let changes save automatically">
    Tone changes save as you release each slider — there is no separate Save button for this section.
  </Step>
</Steps>

## How to connect a knowledge base

<Steps>
  <Step title="Add a source">
    In the **Knowledge base** section of the selected profile, click **+ Source** and add a docs site URL, a Notion space, or a PDF file. Sally may only quote from sources you connect here.
  </Step>

  <Step title="Check the index status">
    Each source shows its chunk count and content type. A green **Fresh** badge means it is indexed and ready. An amber **Re-index** badge means the content has gone stale since the last index.
  </Step>

  <Step title="Re-index when content changes">
    Click an amber badge to re-sync a source. Re-index whenever you publish new docs or update a PDF so Sally always quotes the latest version.
  </Step>
</Steps>

<Note>You can also add **Reference replies** — real prospect question-and-answer pairs Sally studies to learn your voice — and **Use cases & collateral docs**. Uploaded collateral documents have their text extracted and appended to the profile's sales pitch context.</Note>

## How to control auto-send

Auto-send controls when Sally can send a reply without you reviewing the draft first. By default, she auto-sends low-risk reply types and always drafts higher-stakes ones for your approval.

<Steps>
  <Step title="Open the Auto-send panel">
    The auto-send controls are in the right-hand column of the selected profile. Each reply type has an individual toggle.
  </Step>

  <Step title="Configure each reply type">
    Set auto-send on or off per type based on your confidence level:

    | Reply type                                               | Default                |
    | -------------------------------------------------------- | ---------------------- |
    | **Greeting replies** — short acknowledgements and thanks | On                     |
    | **Logistics** — scheduling, links, vCards                | On                     |
    | **Objection handling**                                   | Off (draft for review) |
    | **Pricing questions**                                    | Off (draft for review) |
  </Step>
</Steps>

<Warning>Leave **Objection handling** and **Pricing questions** off auto-send unless you are highly confident in Sally's drafts. With auto-send off, Sally still writes the reply — she simply waits for you to approve it before sending.</Warning>

## How to tune the runtime policy

The runtime policy controls what happens to the *sequence* when a reply is classified — this is separate from the draft Sally writes. There are six handlers, each with an enable toggle and a one-line config summary.

| Handler                         | Default auto-send | What it does                                                          |
| ------------------------------- | ----------------- | --------------------------------------------------------------------- |
| **Positive response**           | Configurable      | Stops or marks the sequence when clear buying intent is detected.     |
| **Out of office reply**         | Configurable      | Pauses the sequence when temporary unavailability is detected.        |
| **Meeting booked**              | Configurable      | Stops outreach and updates status on booking intent.                  |
| **Unsubscribe / stop emailing** | **Always on**     | Hard compliance stop — cannot be disabled.                            |
| **Email bounce**                | **Always on**     | Stops automation on permanent delivery failures — cannot be disabled. |
| **Not interested**              | Configurable      | Stops this sequence without globally unsubscribing the prospect.      |

<Warning>The **Unsubscribe / stop emailing** and **Email bounce** handlers are always-on and cannot be disabled. These protect your sending reputation and ensure compliance with opt-out requirements.</Warning>

<Steps>
  <Step title="Review the six handlers">
    Scroll to the **Runtime policy** section of the selected profile. Each handler shows its name, a short description of when it triggers, and an enable/disable toggle.
  </Step>

  <Step title="Enable or disable inline">
    Flip each handler's toggle directly in the list. The two compliance handlers are marked **Always on** and cannot be toggled off.
  </Step>

  <Step title="Edit the handler details">
    Click **Edit details** on any handler to open its configuration drawer. Set the keywords or sentiment threshold, the action to take (stop / pause / mark interested / global unsubscribe), the notification target, status update, pause duration, and resume behavior.
  </Step>
</Steps>

## How to set topics Sally should never answer

<Steps>
  <Step title="Find the Never reply about section">
    It is in the right-hand column of the selected profile. Click **+ add**, type a topic Sally should refuse to draft about, and press Enter. Remove an entry by clicking the **×** on its chip.
  </Step>

  <Step title="Understand the behavior">
    When an incoming reply matches one of these phrases, Sally does not draft a response. Instead, she routes the thread to your review channel and leaves it untouched for you to handle manually.
  </Step>
</Steps>

## Tracking 30-day performance

The **Performance** section of each handler profile shows 30-day metrics for that profile:

* **Drafts written** — total replies Sally drafted.
* **Auto-sent** — replies sent without human review.
* **Edited** — drafts you modified before sending.
* **Reply-to-meeting** — the rate at which handled replies converted to a booked meeting.

Use these numbers to calibrate auto-send settings and tone sliders over time.

## FAQ

<AccordionGroup>
  <Accordion title="What's the difference between Live and Draft?">
    **Live** means the profile is active and Sally uses it on incoming replies. **Draft** means the profile is saved but inactive — use it to build or revise a profile without affecting live conversations.
  </Accordion>

  <Accordion title="How does Sally choose which profile to use?">
    Sally reads the incoming reply and picks the best-matching profile. If no profile is a clear match, she falls back to the account **default** profile — so it is worth keeping a well-configured default active at all times.
  </Accordion>

  <Accordion title="Why is Sally drafting instead of sending automatically?">
    That reply type has auto-send turned off — for example, pricing or objection handling — or the reply matched a **Never reply about** topic. Check the Auto-send panel and your hard-stop phrases for the active profile.
  </Accordion>

  <Accordion title="Can I prevent Sally from quoting specific information?">
    Yes. Only sources connected in the **Knowledge base** section are quotable. Remove a source to stop Sally referencing it. Use **Never reply about** entries for topics she should never address at all, regardless of what is in the knowledge base.
  </Accordion>

  <Accordion title="A knowledge source shows an amber badge — what does that mean?">
    The source is stale — its content has changed since the last index, or initial indexing has not completed. Click **Re-index** to re-sync it so Sally quotes the current version.
  </Accordion>
</AccordionGroup>

## Related

<CardGroup cols={2}>
  <Card title="Company Profiles" icon="building" href="/ai-library/company-profiles">
    Give Sally the business context she draws on when she replies.
  </Card>

  <Card title="Persona Profiles" icon="user" href="/ai-library/persona-profiles">
    Define who you sell to so replies land for the right audience.
  </Card>

  <Card title="Prompt Library" icon="scroll" href="/ai-library/prompt-library">
    Reusable prompts that shape how Sally writes across every flow.
  </Card>

  <Card title="AI SDR Builder" icon="robot" href="/ai-agents/ai-sdr-builder">
    Pick a reply handler when you build an automated outbound sequence.
  </Card>
</CardGroup>