> ## 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.

# Connected AI vs. a generic chatbot

> What changes when the AI can reach Hillspire's real systems — and why Aperium exists.

Most people already have ChatGPT or Claude open in a browser tab. Both are genuinely good tools. So why does Hillspire have its own AI platform? The answer is one word: **connection**. A generic chatbot can only talk. Aperium can reach Hillspire's real systems and act on real data — and that difference is the entire reason the platform exists.

## Generic chatbots are genuinely useful

Give credit where it is due. Public tools like ChatGPT and Claude are excellent at writing, editing, brainstorming, explaining a concept, and drafting a first pass at almost anything. Aperium does all of that too. That capability is table stakes, not the differentiator — for that category of task, "just use a chatbot" is often the right call.

## The wall a chatbot can't get past

No matter how capable the underlying model is, a generic chatbot has never seen your deal record, your Jira ticket, the general ledger, or a Slack thread. It only knows what you personally type into the box.

<Warning>
  A generic chatbot **cannot look anything up** in a Hillspire system, and it **cannot take an action** anywhere — because it isn't connected to anything of ours. If you didn't paste it in, the model is guessing.
</Warning>

## Guessing vs. knowing: two examples

<CardGroup cols={2}>
  <Card title="Drafting a deal email">
    A generic chatbot writes a nice-sounding email about "the deal" from whatever details you happened to type — or invents plausible ones if you didn't. Aperium pulls the actual investment record and drafts the email from the real terms, numbers, and status.
  </Card>

  <Card title="A project status update">
    A generic chatbot produces plausible phrasing with no way to know what's really open or blocked. Aperium reads the actual tickets or project data and reports what's genuinely done, blocked, or overdue — as of today, not as of whenever the model's training ended.
  </Card>
</CardGroup>

That gap — a well-written guess versus an answer grounded in the real record — is the whole point.

## The systems Aperium can reach

These are real, live connections, not a roadmap. This is a high-level tour; the [FAQ](/using/faq) and the connections catalog go deeper.

| System           | What it covers                |
| ---------------- | ----------------------------- |
| Salesforce       | CRM and deal records          |
| NetSuite         | Accounting                    |
| Malbek           | Contracts                     |
| Atlassian        | Jira and Confluence           |
| Google Workspace | Email, calendar, Drive        |
| Microsoft 365    | Mail and documents            |
| Slack            | Messages and channels         |
| Odoo             | Business operations           |
| Procore          | Construction and project data |
| BigQuery         | The company data warehouse    |

## Why it was built this way

Public chatbots are walled off from company systems by design. That is good for security — but it permanently caps what they can ever do for you. Aperium exists specifically to be the secure bridge: AI running **inside** Hillspire's own environment, with scoped, permissioned access to our real systems.

<Info>
  That bridge is not a feature bolted onto Aperium after the fact. It **is** the reason the platform exists.
</Info>

The same logic explains why you still shouldn't paste sensitive data into a public tool: the public tool is outside our walls, and Aperium is the option that lives inside them.

## What this changes for you

<CardGroup cols={3}>
  <Card title="Fewer round-trips" icon="rotate">
    Fewer "let me check and get back to you" loops — the AI can just check.
  </Card>

  <Card title="Grounded answers" icon="anchor">
    Responses come from the actual record instead of a plausible guess.
  </Card>

  <Card title="It can act" icon="bolt">
    You can ask it to take the next step, not only describe one.
  </Card>
</CardGroup>

<Tip>
  The one line to remember: a generic chatbot can write *about* your work. Aperium can reach it — and act on it — because it's actually connected to the real thing.
</Tip>

## Next

Ready to sign in and run your first request? Head to [Getting started](/using/getting-started).