> ## 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.
# Features deep dive
> Tools & integrations, the model picker, scheduled automations, and Artifacts.
Getting into Aperium for a single session is only the start. Most of the value shows up once Aperium is part of your regular week — reaching real systems on demand, delivering recurring answers automatically, and saving results you can return to and share. This page tours the features that make that possible.
## Tools & integrations
A tool or integration is a specific connected system Aperium can reach on your behalf mid-conversation — Salesforce, Jira, Procore, and others. Reach for one any time your answer needs to depend on a real, current record rather than general knowledge: "what does our Jira board actually say," not "what does a project status update usually look like."
**Example.** Ask "Pull the open tickets on this Jira project and tell me which are blocked" instead of drafting a status update from memory. Watch for the tell — a short pause or a visible action step before the answer — and then check the result against the source system.
### Risk levels: read-first
Every tool a connection exposes carries a risk level:
* **Read-only** access just looks things up — lower risk.
* **Write** access means Aperium can change something in that system — higher risk.
Grant read-only first, and upgrade to write later once you're comfortable
with how a tool behaves. It pairs with the "verify, don't blindly trust"
habit from [Responsible & safe use](/using/responsible-use).
If a system you need isn't connected yet, ask the Aperium team — it can
usually be added. See [What you can connect to](/using/connections).
## Scheduled automations
Instead of asking the same question every week, set Aperium up to run it on a schedule and deliver the result to you automatically. Use one for anything you ask on a repeating cadence — daily, weekly, monthly — where only the underlying data changes, not the question. This turns Aperium from something you remember to ask into something that has already done the work by the time you sit down.
A few cadences teams use, one system each:
| Cadence | System | Job |
| ------- | ---------- | ------------------------------------------------------------------------------------- |
| Weekly | Malbek | "Stuck contracts" — contracts sitting in Draft, Legal, or awaiting signature too long |
| Friday | Gmail | "Loose ends" — emails from this week you never replied to |
| Monday | Salesforce | "Weekly account review" — accounts and opportunities stalled or closing soon |
| Daily | Jira | "My open work" — tickets assigned to you, plus anything overdue |
### Setting one up
Begin with a chat or prompt you've already run successfully — for example,
the "which contracts are stuck" ask.
Instead of re-asking, save the working prompt as a recurring job.
Choose how often it runs — weekly fits the stuck-contracts example.
Results arrive in-platform or as a chat message today, with Slack delivery
available where enabled.
Scheduled jobs run on Aperium's own always-on servers, not your laptop, and
you can review past runs any time.
## Sessions & history
A new session is a fresh page, not an erasing of the old one — past sessions aren't gone. You can typically return to a prior conversation and keep going, or find an answer from last week without re-asking from scratch. Reach for this when you're continuing work on the same topic — a deal, a project, a ticket — rather than starting something new.
## Sharing & reusing results
A good result, or a request that worked well, doesn't have to live only in your own conversation. Sharing what Aperium produced — or the prompt that produced it — saves a teammate from starting at zero. Building a durable library of prompts and workflows worth reusing is covered in [Build & share your own workflows](/using/workflows).
## Artifacts
A dashboard or report Aperium builds doesn't have to end when the session does. Save it as an **Artifact** — a persistent object you can share like a Google Doc, so a teammate can open it directly without rebuilding your work. Two things make it more than a static export:
* **Ask follow-up questions of the data** right inside it — "break this down by region," "what changed since last quarter."
* **Reshape the chart in plain English** — "make it a stacked bar," "change the colors" — without opening a BI tool.
Reshaping charts in plain English is an in-progress capability — it works,
but it's still maturing, so set expectations accordingly.
A contract-spend dashboard or a help-desk-tickets dashboard that started as a one-off question can live on this way as something the team returns to and shares.
## Model picker
Aperium lets you pick which AI model handles a task, and you can switch on the fly mid-task. The choice comes in three plain-language tiers:
| Tier | Best for |
| --------- | -------------------------------------------------------- |
| Light | Quick, simple, one-off asks |
| Mid-range | Daily work — the sensible default |
| Frontier | Large, complex, multi-file jobs that need deep reasoning |
The picker offers both the Anthropic and OpenAI model families. Most days the mid-range default is right — reach for frontier when the job justifies it, and light when you just need something fast.
## Before you rely on it
* A scheduled job is only as good as the access behind it. If you lose access to a system, the job can quietly keep running and stop returning useful results — apply the same "verify, don't blindly trust" habit from [Responsible & safe use](/using/responsible-use).
* Before sharing a result, confirm who else can see it — the same way you'd think twice before forwarding any other document.