Skip to main content

Use cases

Build the extension your business needs.

The base is built in: AI agents, automation, local models and document search. Over a documented interface you add exactly what is missing.

Compact AI machine running Ara OS on a desk, with a monitor showing an open editor behind it
Ara OS runs on your own box

Out of the box

The base is complete

This runs on the device from day one. You build on top of it, instead of starting from scratch.

ara-osls /ara/base
agents/Configure AI agents for fixed tasks. Each one gets tools, context and a model.
n8n/Automate workflows and trigger your agents, for example by message or on a schedule.
documents/Search contracts and manuals. The hits become knowledge you hand to an agent.
models/Local language models via Ollama, plus embeddings for the document search.
api/Everything is reachable through a documented, local interface.
ara-os

Build on top

What you can build on top

A few concrete examples. Open them up. What is possible is decided by your needs, not by a catalog.

Quote assistant for trades

Situation

Quotes get written in the evening, once the job site is done. Two hours of typing from the measurements, then looking up prices again. Whoever answers faster wins the job.

How it runs

  1. 01The fitter photographs the measurements and drops them in a folder on the device.
  2. 02An agent reads out the line items, quantities and materials included.
  3. 03It calculates with your own price list, not with list prices from the web.
  4. 04The finished quote lands as a PDF in the customer folder, for review before it goes out.

What it gives you

The quote goes out the same day instead of on Friday. Your pricing stays on the device in the workshop and goes to no provider.

Requests from the inbox

Situation

The shared inbox gets everything: real requests, invoices, ads, reschedules. Someone has to sort it every morning, and the urgent request gets lost in the pile.

How it runs

  1. 01Ara OS pulls the mail from the inbox you already use.
  2. 02An agent sorts by type: request, invoice, appointment, rest.
  3. 03For each request it searches your records for whether the customer is known.
  4. 04Standard cases get a draft reply, the rest is routed to the right person.
  5. 05Nothing goes out that someone has not approved.

What it gives you

The inbox is sorted in the morning instead of full. The mail traffic stays in the building, because the model computes locally.

Deadline checks for law firms

Situation

Notice periods sit somewhere on page eleven. Miss one and you are liable. So someone reads them by hand, on every new contract.

How it runs

  1. 01New contracts land in a watched folder.
  2. 02A workflow searches them for deadlines, terms and renewal clauses.
  3. 03Every find is noted with its source location, so you can double-check it.
  4. 04The dates move into the firm calendar, with lead time for a reminder.

What it gives you

No more deadline that lives in one person's head. Built once, the same extension runs in every firm.

Invoice and receipt processing

Situation

Receipts arrive as a photo, as a PDF, as a scrap of paper in a shoebox. Accounting types them up, and half of them are still missing at month end.

How it runs

  1. 01Receipts get photographed or dropped in by mail.
  2. 02An agent reads out amount, date, tax rate and supplier.
  3. 03It reconciles against open items and flags what does not match.
  4. 04Outgoing invoices are created directly as ZUGFeRD, so machine-readable.

What it gives you

Accounting gets clean data instead of a stack. Revenue and customer data stay local on the device.

This gets built by you or your partner

An extension is in the end a single file that you drop into Ara. The full docs live in the signed-in area.