> ## 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.
# The agent catalog
> Tour the available agents and choose the right one for the job.
Aperium isn't one do-everything agent. It's a small team of specialized agents, each carrying a different set of tools for a different domain. Picking the right one is often the difference between a good answer and a great one — ask for "this quarter's pipeline" from the wrong specialist and you may get a well-written paragraph instead of the chart you wanted.
## Why several agents instead of one
You wouldn't ask the receptionist to audit the books — not because they couldn't learn it, but because someone who lives in that system every day gives a faster, more precise answer. The same logic applies here. An agent scoped to fewer systems can be more focused and reliable in its domain than one agent trying to reach everything at once.
Think of an agent as a session pre-loaded with the context and data sources a role needs. Someone working in a recruiting agent doesn't re-explain what data exists, what's missing, or where to find it every session — the agent already carries that context.
## The agents
The agents below are known examples, not a confirmed complete roster. Your
live agent picker in Aperium is the source of truth for the current list,
the exact labels, and what each agent can reach. Check it before relying on a
single agent for a task.
Open-ended questions, writing, and brainstorming. Your default when nothing
more specific fits.
CRM work — accounts, opportunities, and contacts.
Charts and visual summaries of data.
What's new or recently changed — timely, fast-moving information.
## Agents call skills automatically
While you're chatting with an agent, it can reach for a **skill** — a saved, repeatable routine — on its own, mid-answer, without you invoking it by hand. A couple of examples seen in practice:
* A NetSuite lookup skill: "when someone asks about customers or sales, here's how to find it" — the same lookup every time, no improvising.
* A multi-step "total contract spend by vendor" skill: pull the vendor's contracts, pull each contract's invoices, and sum them — three steps, one saved routine, one number back.
The distinction: the **agent** is the context you pick to start a session; the **skill** is a routine running underneath it, the same way every time, no matter who's asking.
## How to choose
| What you're asking | Start with |
| --------------------------------- | ---------- |
| Write an email or brainstorm | General |
| Look up an account or opportunity | Salesforce |
| Show a chart or visual summary | Dashboard |
| What's changed recently | Recency |
| Not sure, or a mixed task | General |
When in doubt, start with **General**. It's the safe default and can often
point you toward a more specific agent for the job.
## Switching agents
If an answer feels generic or off-target, that's your cue to check whether a more specific agent exists for the job — regardless of how switching works in your version of the UI.
## Common mistakes
* **Assuming an agent knows about a system it isn't connected to.** It can only reach what it's wired to.
* **Sticking with the wrong agent** long after the answers stop being useful.
* **Assuming a specialized agent replaces your judgment.** Specialization doesn't remove the need to verify outputs — see [Responsible & safe use](/using/responsible-use).