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

# What the assistant can do

> A plain-language catalog of what you can ask the Custral AI assistant to do, with example prompts for each capability.

The Custral assistant isn't just a chatbot. It can **act across your whole
workspace**. Ask in plain language and it will find things, draft and send
messages, create records and pages, and even build a slide deck. Anything that
*changes* your data asks for your approval first (see [Permissions](/ai/permissions)).

<Tip>
  Open the assistant anywhere with **⌘I** (or the spark icon in the sidebar).
  Type what you want in your own words. You don't need to name the tool.
</Tip>

## Find & answer

Search across records, conversations, and email, and get a straight answer or a
summary.

<CardGroup cols={2}>
  <Card title="Search your workspace" icon="magnifying-glass">
    Find records, conversations, and emails by anything on them.
    <br />**Try:** *"Find every open deal over \$10k owned by me."*
  </Card>

  <Card title="Summarize" icon="list">
    Condense a record, a thread, or a set of results.
    <br />**Try:** *"Summarize my conversation with Acme this week."*
  </Card>

  <Card title="Research the web" icon="globe">
    Search and read public pages to enrich an answer.
    <br />**Try:** *"What does this company do? research acme.com"*
  </Card>

  <Card title="Answer from your data" icon="circle-question">
    Ask questions grounded in your records.
    <br />**Try:** *"Which contacts haven't been touched in 30 days?"*
  </Card>
</CardGroup>

## Records & data

Create and update records, one at a time or in bulk.

<CardGroup cols={2}>
  <Card title="Create records" icon="plus">
    Add a company, contact, deal, task, or anything you've modeled.
    <br />**Try:** *"Create a deal for Acme worth \$25k in the Negotiation stage."*
  </Card>

  <Card title="Update in bulk" icon="pen-to-square">
    Set a value across many records at once.
    <br />**Try:** *"Mark all deals closed before June as Archived."*
  </Card>
</CardGroup>

## Build & structure your workspace

Model new data and lay out the surfaces to work with it.

<CardGroup cols={2}>
  <Card title="Design objects & fields" icon="table-cells">
    Create objects and properties, or a whole related schema at once.
    <br />**Try:** *"Add a 'Renewals' object with fields for owner, ARR, and renewal date."*
  </Card>

  <Card title="Views, pages & dashboards" icon="grip">
    Build tables, kanban boards, and dashboards; add filters.
    <br />**Try:** *"Make a kanban of deals grouped by stage on a new page."*
  </Card>
</CardGroup>

## Email & conversations

Work your inbox without leaving the chat.

<CardGroup cols={2}>
  <Card title="Draft & send email" icon="envelope">
    Compose a reply in your voice; send it (with your OK).
    <br />**Try:** *"Draft a follow-up to Jane thanking her for the call."*
  </Card>

  <Card title="Act on conversations" icon="comments">
    Look up a thread and complete its pending actions.
    <br />**Try:** *"What's the next action on the Globex conversation?"*
  </Card>
</CardGroup>

## Calendar & tasks

<CardGroup cols={2}>
  <Card title="Schedule & check availability" icon="calendar">
    Read events and free/busy, and create meetings.
    <br />**Try:** *"Book a 30-min call with Acme next Tuesday afternoon."*
  </Card>

  <Card title="Tasks & notes" icon="circle-check">
    Create, complete, and list tasks; jot a note on a record.
    <br />**Try:** *"Add a task to send Acme the proposal by Friday."*
  </Card>
</CardGroup>

## Automate

<CardGroup cols={2}>
  <Card title="Workflows & sequences" icon="bolt">
    Inspect automations and enroll records into outreach sequences.
    <br />**Try:** *"Enroll these 12 new leads in the Outbound sequence."*
  </Card>

  <Card title="Skills & /commands" icon="wand-magic-sparkles">
    Run reusable procedures your team has saved. A saved
    [Skill](/ai/rules-and-skills) with a command shows up as a **/command**.
    <br />**Try:** *"/qualify-lead"* or *"use the lead-qualification skill."*
  </Card>

  <Card title="Build a playbook" icon="list-check">
    Draft a [playbook](/comms/playbooks/overview). The steps a rep should
    follow on a call, and list or inspect the ones you already have.
    <br />**Try:** *"Create a discovery-call playbook for inbound demos."*
  </Card>

  <Card title="Create a scorecard" icon="clipboard-check">
    Build a [coaching scorecard](/comms/coaching/scorecards) so calls get
    graded against criteria you choose.
    <br />**Try:** *"Make a scorecard that grades discovery calls on rapport, qualification, and next steps."*
  </Card>
</CardGroup>

## Documents & presentations

<CardGroup cols={2}>
  <Card title="Draft documents" icon="file-lines">
    Create rich workspace documents from a brief.
    <br />**Try:** *"Write an onboarding doc for the Acme account."*
  </Card>

  <Card title="Build a slide deck" icon="presentation-screen">
    Generate a **fully branded** presentation from a topic, the assistant
    writes the slides and Custral applies your brand theme and layout, then
    hands back a link.
    <br />**Try:** *"Make a 10-slide customer pitch deck for Acme."*
  </Card>
</CardGroup>

<Note>
  New: **"Make me a deck"** creates a real
  [Presentation](/blocks/presentations/overview) in your workspace, themed and
  ready to present or edit.
</Note>

## Before it acts

Reading is always free. Anything that **changes data or reaches outside
Custral** (creating a record, sending an email, building a deck) pauses for
your approval, so nothing happens behind your back. You can loosen or tighten
this in the assistant's mode. See [Permissions](/ai/permissions).

Each [chat agent](/ai/chat-agents/overview) can also be limited to a chosen set
of tools (and kept off the web), so a focused agent only offers what it should.

## Troubleshooting

<AccordionGroup>
  <Accordion title="The assistant says it can't do that / doesn't have that tool">
    The agent you're talking to may be **scoped** to a subset of tools, or have
    web access turned off. Switch to a less-restricted agent, or widen the
    agent's tools under [Tool gating](/ai/chat-agents/tools).
  </Accordion>

  <Accordion title="It keeps asking me to approve every step">
    Writes are gated by design. If you trust a run, approve it, or change the
    assistant's mode so writes auto-run. Destructive actions (like deleting)
    always ask. See [Permissions](/ai/permissions).
  </Accordion>

  <Accordion title="It built a deck but I can't find it">
    The assistant returns a **link** to the new deck and saves it as a
    Presentation page. Open the link, then drag it into a sidebar section to
    file it wherever you like. Everything on the deck is editable, and you can
    hit **Present** to show it.
  </Accordion>

  <Accordion title="The result isn't quite what I wanted">
    Give it more detail and ask again, e.g. the audience, the key points, or
    the number of slides for a deck. Drafts are never sent or published without
    you, so it's safe to iterate.
  </Accordion>
</AccordionGroup>

## Related

<CardGroup cols={2}>
  <Card title="Tools overview" icon="screwdriver-wrench" href="/ai/tools/overview">
    The tool groups behind these capabilities.
  </Card>

  <Card title="Chat agents" icon="robot" href="/ai/chat-agents/overview">
    Build focused personas with their own tools and scope.
  </Card>

  <Card title="Rules & skills" icon="book" href="/ai/rules-and-skills">
    Teach the assistant your process and save reusable /commands.
  </Card>

  <Card title="Permissions" icon="shield-check" href="/ai/permissions">
    How the approval gate protects your data.
  </Card>
</CardGroup>


## Related topics

- [AI](/ai/overview.md)
- [Presentation](/data/properties/communication/presentation.md)
- [Get started with AI](/ai/getting-started.md)
- [Email](/comms/email/overview.md)
- [Blocks](/blocks/overview.md)