AI automation: hand the repetitive work to software.
AI automation means handing the repetitive manual work in your business over to software: reading incoming messages and opening records, pulling data from invoices, summarising reports, assigning tasks by rule. At Pirdir, automation runs on top of your records and by your rules.
Who it’s for
Teams processing messages, documents and forms
Where it runs
WhatsApp, email, forms, documents
Setup
2–6 weeks
Starting price
Module from €650
Today every message is read and every invoice typed in by hand. In Pirdir, software takes the first step.
Topic
Document
Assignment
Record
Summary
The boiler is dripping again
Who is this email for?
How the assistant handles a message
As you scroll, the assistant works through the message step by step.
It reads
The text of a message, email or document is read.
It understands
The topic, urgency and related product are worked out.
It matches
The registered customer and related records are found.
It opens a record
A record is opened by rule, or sent for approval.
A Monday in a heating firm's office
A day at a heating and plumbing company with nine vans, where software reads messages and receipts first. The team still makes the final call.
08:15
Weekend messages already sorted
37 messages came in by WhatsApp and email over the weekend. The assistant has sorted 22 as faults, 9 as quote requests and 6 as other; the faults are waiting on the service list.
09:00
The weekly summary is ready
Tom, the manager, gets last week's summary: 64 jobs closed, and 3 customers wrote about the same fault a second time. All three are named in it.
10:20
A boiler message becomes a job
Emma Novak writes that her boiler is dripping again. The assistant finds her and the unit, sees the visit two months ago and opens T-1093 for the service team.
11:05
A fuel receipt in 20 seconds
Marco photographs his fuel receipt. Supplier, date, amount and VAT are read; the category stops at 71%, so the receipt goes to Sofia, the accountant, for approval.
15:30
A supplier invoice is matched
12 fields are read from a scanned invoice that came by email, and its lines are matched to the purchase order. One quantity does not match, so the invoice reaches Sofia flagged.
How we work
Before any code is written, we look together at what we’ll build, how long it will take and what it will cost.
Example project: automation that reads messages and receipts
The price for this project was itemised in writing after discovery. The timings belong to the example company; yours are set out in discovery.
Discovery
200 messages and 60 receipts from the last three months show which jobs are a good fit for automation.
Setup
Message sorting, receipt reading and invoice matching are set up; every result goes to the approval screen.
Working with approval
The office and accounts approve every item; fields read wrongly are noted and fixed.
Rules
Fields above 95% accuracy move to working without approval; the rest stay in approval.
Heating and plumbing service · 9 vans, 3 office staff
Which AI service is used and how long data is kept were written into the quote. Amounts in euros, excluding VAT.
What AI and automation cover
Everything below comes ready. Field names, statuses and reports are adjusted to your business during setup.
Message classification
Topic, language and urgency.
Document reading
Extracting fields from invoices, delivery notes and forms.
Opening records
Service tickets, orders, tasks.
Summaries and reports
Weekly summaries and situations that need attention.
Rule-based automation
Assignment, reminder and approval flows.
Human approval
Anything uncertain goes to the team.
11:05 · Service van
Marco photographs the receipt; the amount and VAT are read, and the uncertain category goes to Sofia for approval.
Who is AI automation for?
It isn’t for everyone. We also spell out when it won’t fit, so nobody wastes time on a call.
A good fit
You may not need it
Set up alongside it
They connect to the same core and share customers, documents and users.
Let’s hand repetitive work to software.
On the call we list the jobs your team does by hand every day, and show with examples which of them can be automated.
AI automation: common questions
Ask anything not covered here on the call, and we’ll send the answer in writing too.
What if the AI makes a mistake?
At first, actions go to the team for approval. As you see how accurate it is, you decide which jobs it can do without approval, and every action is logged.
Does our data go to an AI company?
The text being processed is sent to an AI service. The choice of service, the retention terms and your GDPR privacy notice are agreed together during setup.
Which jobs can be automated?
Jobs whose rules can be described and whose data is recorded: classifying messages, reading documents, opening records, reminders, assignments, reports.
Does it work well in languages other than English?
Yes. Messages and documents in German, Turkish and other common languages are processed, and typos and abbreviations are mostly understood.
How is the price of automation worked out?
By the scope of the work. The AI assistant starts at €650 as a module; custom automations are quoted after discovery.
Let’s hear about your business in 20 minutes.
The call is free. By the end you’ll know which route the project should take, how long it will take and how discovery will go.
On the call you’ll speak with Anıl and Hatice. They are also the ones who build your software.