An Irish solicitor practice with three fee-earners was spending fifteen to twenty hours every month generating Legal Aid forms (LA1, LA10, CLA11, CF1) by hand, from case data that already existed in their own practice system. After business process automation was applied to that single task, the same forms were produced in under two minutes. That is business process automation: not a concept, not a software category, not a future investment. Something that is running in an Irish practice right now.
Here is the definition, once: business process automation is the use of technology to handle repeatable, rule-based tasks that currently require human effort. It does not replace your core systems. It connects them, so the manual step in the middle stops existing.
The rest of this piece is examples, numbers, and a way to work out which processes in your own business are worth automating.
What makes a process worth automating
Not every process should be automated. Automating something that is highly variable or badly defined adds complexity and creates a system that breaks the first time reality changes. Before anything else, a process needs three things in common.
It repeats
The process runs on a regular cycle: daily, weekly, or every month-end. One-off work has nothing to amortise the build against.
It follows rules
The steps can be written down without ambiguity. If two people in the business would do it differently, it is not ready yet.
It costs something measurable
The manual version has a price in hours, in errors, or in delayed cash. If you cannot measure it, you cannot prove the payback.
Processes that fit comfortably: reconciling invoices between an accounting package and a CRM, generating standard legal or compliance documents from records you already hold, and converting logged work-in-progress hours into draft client bills.
Processes that do not fit: negotiating commercial terms with a client, and resolving a non-standard customer complaint. Both need judgment, context, and the ability to read a room. Automating them produces something worse than the manual version.
The honest test is whether you could hand the process to a new employee with written instructions and expect it done correctly. If you could, software can do it. If you could not, the process is not the problem — the definition is.
Four Irish business process automation examples with real numbers
Each of these is a live system. Each follows the same shape: the manual process that existed before, what was automated, and the outcome that can be measured.
Invoice reconciliation between Salesforce and Xero
Salesforce and Xero do not talk to each other natively. An Irish business was spending four full days every month reconciling invoices across the two systems by hand — partial payments, credit notes, and Relevant Contracts Tax treatment all worked through manually by a senior accountant. After automation, transactions match across both platforms automatically and exceptions are the only thing a human sees.
4 days of senior accounting time, every month
Matching, partial payments, credit notes, RCT treatment
4 days recovered every month
Legal Aid form generation from existing case data
A three-fee-earner practice was spending fifteen to twenty hours a month compiling statutory Legal Aid documentation by hand. The data was already in the practice management system; it was being re-typed into forms. After automation, LA1, LA10, CLA11 and CF1 forms are generated directly from the case record. The fee-earner reviews and signs rather than types.
15–20 hours a month of manual form completion
Form selection, data extraction, fee codes, mileage
Forms generated in under 2 minutes
WIP-to-invoice conversion
Work-in-progress data sitting in a timesheet system is not cash. An Irish professional services firm was converting WIP to invoiced amounts manually, which put a lag between the work being done and the invoice going out. After automation, approved WIP converts to draft invoices automatically and the lag disappears.
Manual timesheet-to-invoice conversion, done in batches
Validated WIP converted to draft invoices on approval
10 days off the cash conversion cycle
Operations coordination across a 30-vehicle fleet
A regional waste operation running thirty vehicles was coordinating billing, missed collections, customer queries, route changes and month-end reconciliation manually, across several people and several systems. Nine specialist AI agents now handle each of those functions. Work that took days runs in hours.
Manual coordination across multiple people and platforms
Nine specialist agents, one per operational function
Multi-day processes now run in hours
The point of the four together is that business process automation is not one thing. It applies to finance, to professional services, to billing, and to operations. The pattern is identical every time: a process that was done by hand is now done automatically, and the result is specific enough to measure. Every one of these is documented in more detail on the case studies page.
What business process automation costs and what it pays back
A single business process automation build sits between €1,000 and €20,000, fixed price, agreed in writing before any work starts. Where it lands inside that band depends on how many systems have to be connected and how clean the data is.
The payback calculation is simple arithmetic: take the cost of running the process manually for a year, and compare it to the build cost. The invoice reconciliation example above recovers four days of senior accounting time a month. At a senior accountant day rate of around €350, that is €1,400 a month, or €16,800 a year, against a mid-band build. Across Keystone's builds the average is a 3× annual saving against implementation cost, with payback inside six months.
If the numbers do not work, the build does not happen. That is what the €1,000 Discovery Report is for — it maps the process, prices the build exactly, and tells you the payback before you commit to anything. There is a fuller breakdown in what AI actually costs for an Irish business.
Business process automation vs AI automation: what is the difference
These get used interchangeably and they are not the same thing.
Traditional BPA
- How it works
- Follows fixed, predefined rules.
- Handles
- Structured data in consistent fields.
- Limitation
- Breaks when the input format changes.
- Example
- Legal Aid form generation — the rules are fixed and the fields never move.
AI-powered BPA
- How it works
- Adds a reasoning layer on top of the rules.
- Handles
- Unstructured inputs — emails, documents, call recordings.
- Limitation
- Needs human review on anything consequential.
- Example
- Email triage — classifies by context and drafts in a specific voice.
Both are valid. The mistake is reaching for the more sophisticated one by default. The right starting point is always the simplest system that solves the problem — if fixed rules do the job, fixed rules are cheaper to build, cheaper to run, and easier to trust. Our AI automation and AI agents pages cover where each one fits.
How to identify which processes in your business are ready
Five questions. Ask them about any process you are considering.
- 01Can you describe it as a series of steps that always happen in the same order?
- 02Does it happen at least once a week?
- 03Could a new employee learn it from written instructions in under an hour?
- 04Is the cost of errors in this process measurable?
- 05Does it currently require someone to move data from one system to another?
If the answer to three or more is yes, the process is a strong automation candidate. If the answer to all five is yes, it is almost certainly the first thing you should build.
Most businesses have more of these than they expect. The reconciliation nobody talks about. The report that gets rebuilt every Monday. The form that gets filled in from data that already exists somewhere else. On a discovery call we go through the list and identify the ones with the highest value, which is usually not the one people expect.
Common questions about business process automation
What is business process automation in simple terms?
It is using software to run the repetitive, rules-based work your team currently does by hand — moving data between systems, generating documents, reconciling records. The work still happens. A person just stops being the one doing it.
What are the challenges of business process automation?
The most common failure is picking the wrong process. Automating something variable or poorly defined creates a fragile system that breaks the first time an input changes, and then costs more to maintain than it saved. The second failure is scope: trying to automate an entire department at once instead of one clearly defined process with a provable payback. Start narrow, prove it, then expand.
Does business process automation mean replacing staff?
Not in any of the builds described above. In each case the recovered time went back into work that only a person can do — client advisory work in the legal practice, review and exception handling in the finance team. What gets removed is the typing, not the role.
Do I need to replace my current systems first?
No. Every example on this page automates work between systems the business already ran — Salesforce, Xero, a practice management system, a timesheet tool. Automation connects what you have. Replacing core systems is a much larger, much riskier project and it is rarely necessary to get the first result.
How long does a business process automation build take?
Around four weeks from kickoff to a system running on your own data. The first build takes longest because it establishes the data layer and the integrations. Anything you add afterwards is faster and cheaper, because that foundation already exists.
The bottom line
Business process automation is not a category of software you buy. It is a decision to stop paying people to move data around. The four examples on this page are unremarkable individually — an accountant's four days, a solicitor's twenty hours, ten days of cash flow, a fleet operation's coordination overhead. Added together across a year they are the difference between a business that scales and one that hires to keep up.
The starting point is one process. Pick the one that fails the five questions least. Work out what it costs you annually. If that number is bigger than the build, you have your first project.
If you would like help identifying which one that is, the Discovery Call is free and takes twenty minutes.