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

# FAQ

> Common questions: getting started, which tool when, models, connections, security, and help.

Answers to the questions people ask most often about Aperium. Don't see yours? Ask in the **#aperium-hillspire-users** Slack channel and we'll add the good ones here.

## Getting started

<AccordionGroup>
  <Accordion title="What is Aperium?">
    Hillspire's own internal AI platform. It does the everyday things you'd expect from ChatGPT or Claude — drafting, summarizing, answering questions — **plus** it can securely reach our real internal systems (email, contracts, CRM, finance, and more) and act on real data, on your behalf. See [Connected AI](/using/connected-ai) for what that changes.
  </Accordion>

  <Accordion title="Do I need to install anything?">
    No. Aperium runs entirely in your web browser.
  </Accordion>

  <Accordion title="How do I get access?">
    Aperium lives at **[hillspire.app.aperium.ai](https://hillspire.app.aperium.ai/)**. Sign in with your Hillspire login and **bookmark it** — Aperium does not appear as an Okta tile today, so the bookmark is your front door. You'll also get the link in your onboarding email.
  </Accordion>

  <Accordion title="What's the first thing I should do?">
    Sign in, follow the short onboarding, and **connect Google first** — it powers the Command Center daily overview (your calendar, emails, and suggested replies). Full walkthrough in [Getting started](/using/getting-started).
  </Accordion>
</AccordionGroup>

## Which tool should I use — Aperium, ChatGPT, or Claude?

<AccordionGroup>
  <Accordion title="What's the default?">
    **Aperium-first.** Use it for everyday chat *and* anything that touches Hillspire's systems or data — that's the thing it can do that the others can't. Reach for **Claude / Claude Code** only for what Aperium can't do yet: working with files on your own computer, writing or running code, and pushing deployments.
  </Accordion>

  <Accordion title="How do I decide in the moment?">
    Two quick questions:

    * **Is it about our systems or data, or just general chat and writing?** → **Aperium**.
    * **Do you need files on your own machine, code you write and run, or a deployment?** → **Claude / Claude Code**, for now.

    For example: "draft an email to a portfolio company" or "find yesterday's Jira sync notes" are Aperium. "Clean up the spreadsheet on my Desktop" or "write a script to reformat these CSVs" are Claude / Claude Code.
  </Accordion>

  <Accordion title="Can I move my custom GPT or saved prompt into Aperium?">
    Yes. A personal custom GPT or saved prompt — say, a "contract reviewer" that flags missing indemnification language or non-standard payment terms — can become an Aperium **agent** or **skill** with the same instructions and judgment. Porting it gets you two upgrades for free: it becomes **shareable** with your whole team, and it can **reuse live connections** — instead of pasting text in, the skill can pull the real document from where it already lives (Drive, Malbek) and act on it.
  </Accordion>
</AccordionGroup>

## Models & cost

<AccordionGroup>
  <Accordion title="Are there usage or token limits?">
    **No limits today.** There may be some in the future, but for now use it freely.
  </Accordion>

  <Accordion title="There are several models — which should I pick?">
    You can switch the model per task. Three tiers, in plain terms:

    | Tier                         | Good for                          | Rule of thumb                                      |
    | ---------------------------- | --------------------------------- | -------------------------------------------------- |
    | **Light**                    | quick, simple, one-off asks       | fast and cheap; fine for easy stuff                |
    | **Mid-range**                | your everyday work                | the sensible default — start here                  |
    | **Frontier** (most powerful) | big, complex, multi-document work | when the answer really matters or spans many files |

    The picker offers the usual Anthropic and OpenAI model families. When in doubt, the mid-range default is the right call.
  </Accordion>
</AccordionGroup>

## Connections — what it can reach

<AccordionGroup>
  <Accordion title="What systems can Aperium connect to?">
    Google Workspace, Microsoft 365, Slack, Salesforce, Malbek (contracts), NetSuite, Odoo, Atlassian (Jira/Confluence), Procore, BigQuery, and more. See [Connected AI](/using/connected-ai) for the full picture.
  </Accordion>

  <Accordion title="How do I connect one?">
    In the onboarding flow or the **Integrations** page, pick a system and authorize it. Start with **read-only** access and upgrade to write access later, once you're comfortable.
  </Accordion>

  <Accordion title="What if the system I need isn't there?">
    Ask the Aperium team in **#aperium-hillspire-users** — it can usually be added.
  </Accordion>
</AccordionGroup>

## Security & permissions

<AccordionGroup>
  <Accordion title="Is my data safe? Does Aperium act as me?">
    Yes. Connections use **OAuth** — in plain terms, *you* sign in to each system yourself, and Aperium then acts **as you**. It can only see and do what your own account is already allowed to. It can't see anything you couldn't see yourself, and your access in each system carries over unchanged.
  </Accordion>

  <Accordion title="What data is OK to put in?">
    Because Aperium is our **internal, connected** platform, it's a safer place for internal work than a public chatbot. Still, verify before you trust or forward anything, and apply the same care you'd give any sensitive data — MNPI, PII, and client or portfolio information — under Hillspire's normal data-handling standards.
  </Accordion>
</AccordionGroup>

## What it can (and can't) do

<AccordionGroup>
  <Accordion title="What is Aperium great at, and where is it weak?">
    **Great at:** drafting, summarizing, extracting, explaining, and — uniquely — pulling from and acting on our real systems. **Weaker at:** precise math and guaranteeing accuracy (always verify figures and quotes). It also can't work with files on your own computer or run code. A human stays accountable for anything you act on.
  </Accordion>
</AccordionGroup>

## The building blocks

<AccordionGroup>
  <Accordion title="Agent, skill, artifact, automation — what do these mean?">
    * **Agent** — a chat session pre-loaded with background and context for a job or role.
    * **Skill** — a saved, reusable routine the AI can follow (and reuse across sessions).
    * **Artifact** — a saved, shareable dashboard or report you can refine in plain English.
    * **Automation** — a task set to run on a schedule (for example, a Monday-morning summary).
    * **Shared / group assets** — agents and skills an admin shares with a whole team; when the owner improves one, everyone gets the newest version automatically.
  </Accordion>
</AccordionGroup>

## Getting help

<AccordionGroup>
  <Accordion title="Where do I get technical support?">
    The **#aperium-hillspire-users** Slack channel.
  </Accordion>

  <Accordion title="Something broke or a session misbehaved — what do I do?">
    Use the **Share** button to share the whole chat thread into the support channel, so someone can see exactly what happened.
  </Accordion>

  <Accordion title="Where do team-specific or operational questions go?">
    Functional working channels are being set up per team.

    <Info>
      The per-team Slack channel plan is not yet confirmed.
    </Info>
  </Accordion>

  <Accordion title="Is there training?">
    Yes — live sessions, recordings, and reference guides, as part of the AI literacy program.
  </Accordion>
</AccordionGroup>

## Accounts

<AccordionGroup>
  <Accordion title="Can I connect more than one account for the same system?">
    Not yet — today it's one connection per system. Multi-account support is in progress.
  </Accordion>
</AccordionGroup>