# What you can do with it

> The concrete version: the applications companies run on it, the data they bring in from elsewhere, and what the AI employees do while everyone else gets on with their work

Source: https://aipril.dev/docs/overview/what-you-can-do-with-it

## Applications, and what they cover

Install one from the catalogue, or describe the one you need. Either way it arrives working. The
list below is a sample; yours will look different.

**Work itself, not reports about it.** Boards with columns you define, cards with an owner, a
deadline and a checklist, dragged along as the work moves. Requests arrive from a form, an email or
a messenger and land in somebody's queue. Recurring work comes back on schedule, and meeting actions
stay on the board. Most companies run this layer in a chat and a spreadsheet.

**Projects, above that layer.** Which projects are running, who is on them, what they cost, what is
late and what is blocked. The portfolio view reads the same tasks people work in, so it stays true
between Friday summaries.

**Sales and customers.** A CRM with the pipeline and the activity. Quotes, with the discount
approval attached. Contracts and their renewal dates. Customer claims and complaints, with an owner
and a clock.

**Buying and suppliers.** Purchase requests and their approvals. Supplier vetting before the first
invoice. Comparing bids. Matching an invoice against what was ordered and what arrived.

**People.** Onboarding, and the offboarding everybody forgets until security asks. Absence and
holidays. Recruitment. Reviews and one-to-ones. Training and certifications, with expiry dates
somebody watches. Expense claims. A portal where all of that lives for the person it belongs to.

**Operations, whatever yours are.** An equipment register with a maintenance schedule. Stock with
batches and shelf life. Quality incidents and the corrective actions that follow. Field requests.
Facilities and office issues. A fleet. Shipments.

**Compliance and risk.** Registers where "approximately complete" is a finding. Internal audits and
what they turn up. Policy sign-off, with a record of who acknowledged what. A risk register with a
review date on it.

**Money, without touching your accounting.** Budgets and what is committed against them. Price
lists and their approvals. Loyalty or bonus points. A view of what a counterparty owes. The ledger
under all of it is real double entry. See [Ledger](https://aipril.dev/docs/ledger.md).

**Marketing.** Campaigns from idea to plan. A content calendar. Events, and everything that has to
happen before one.

**And the registers nobody ever built a system for.** You have four or five: the contracts folder,
the equipment list, the exceptions log, the spreadsheet on one person's laptop. None of them was
ever worth a project. Each of them takes an afternoon here.

See [Solutions](https://aipril.dev/docs/solutions.md) for what is ready to install, and
[Applications](https://aipril.dev/docs/apps.md) for how one is made.

## Your data, from wherever it is

You do not have to move your accounting to get reporting on it.

- **Pull your finances in from QuickBooks**, your ERP, your warehouse system, as a live connection
  or refreshed on a schedule
- **Databases and services** - PostgreSQL, Oracle, MS SQL, ClickHouse, anything with an API
- **The spreadsheets people actually maintain** - uploaded, or collected automatically from a
  folder, a share or a Google Sheet
- **And joined together**, so revenue from one system sits beside costs from another and requests
  from a third

Then build the dashboards over it and share a link, not a screenshot.
See [Analytics](https://aipril.dev/docs/analytics.md).

## The work that has always lived in spreadsheets

Planning, budgets, forecasts, the list somebody maintains by hand. Keep the grid and the formulas
people already know. What they type there is data the rest of the system can use: a plan can start
an approval, and two people can be in the sheet at once.
See [Spreadsheets](https://aipril.dev/docs/spreadsheets.md).

## And the AI employees do the routine

- **They watch.** The deal that has not moved in eleven days. The stock that runs out before the
  next delivery. The request nobody picked up on Friday. The invoice that does not match the order.
- **They check.** Look the company up, fill in what can be filled in, verify a record against
  another system, flag the field that contradicts itself.
- **They remind.** The approval sitting with somebody on holiday, the document nobody attached, the
  deadline on Thursday - in Slack, Telegram or email, where the person actually is.
- **They prepare.** The summary, the draft reply, the figures for the decision, so the decision
  takes two minutes.
- **They signal.** Anything outside what they may handle goes to a person, with what they found.

They work inside your applications, under the same permissions as everyone else, and everything they
do is in the same history. See [AI organisation](https://aipril.dev/docs/overview/ai-organisation.md).

## Where to start

Pick one process that annoys everybody: the approval that goes missing, the report somebody rebuilds
every month, the register kept in a file. Install the closest ready application or describe what you
need, run it with real work for a fortnight, and change what turns out to be wrong.
See [Getting started](https://aipril.dev/docs/overview/getting-started.md).