Repeatability
A fixed sequence of steps and rules that can be written down. If expert judgement decides every single case, the process is not a candidate.
PROCESS AUTOMATION
A catalogue of 53 processes described step by step, assessment criteria and a two-minute test showing whether a given process is a good candidate. Based on deployments in manufacturing, trade and public institutions.
CHOOSING THE FIRST PROCESS
We recommend starting with a single, moderately complex process whose effect is visible within weeks. A process that is too simple convinces nobody, while one that is too complex stretches the rollout over months and drains the team's enthusiasm. Only after the first deployment is it worth surveying the whole organisation and sequencing what comes next.
A fixed sequence of steps and rules that can be written down. If expert judgement decides every single case, the process is not a candidate.
The biggest gains come from monotonous, frequent work. The threshold where automation pays back quickly is usually a dozen or more hours a month.
Systems, files, portals, e-mail. Paper-only documents must first enter a digital flow, otherwise there is nothing for the automation to process.
Systems that will not be replaced within the next year. When a replacement is planned, base the automation on what will not change: institutional portals and official formats.
Exceptions do not rule automation out, provided they can be detected unambiguously. The automation handles typical cases and routes the rest to a person with the reason stated.
Someone who can describe the flow and will be available during analysis and testing. This is one of the most common reasons deployments slip.
QUALIFICATION TEST
Think of one specific process and answer ten questions. The score is not an oracle, but in practice it shows well whether this process is worth discussing first.
Answered: 0 / 10
PROCESS CATALOGUE
The list comes from deployments in organisations of different sizes. For each process we state what the automation actually does, in which systems and how often. Items marked as an obligation stem from regulations in force in 2026.
The automation downloads incoming invoices from KSeF, stores the XML file together with a PDF rendering in the document repository and identifies the supplier by tax number. Based on posting history it proposes a cost account and cost centre, then routes the document into the right approval path. Invoices without a clear match go to a list for an accountant, with the reason stated.
The automation reconciles the sales ledger with official confirmations and flags documents that received no KSeF number or came back rejected. Where the cause was technical it resends; substantive errors go back to the issuer with the system message attached. Each day closes with a ledger-to-system compliance report.
The automation matches purchase invoice lines against the order and the goods receipt note: item code, quantity, unit price and delivery terms. Lines that agree within the agreed tolerance go to posting; differing ones are described line by line: a price other than the order, a quantity that does not match the receipt, an invoice with no receipt or a receipt with no invoice. Each case goes to the buyer who owns the order rather than to a shared mailbox.
For services, rentals, transport, licences and utilities there is no goods receipt, so the automation matches the invoice straight to the order or the contract: the rate, the billing period, the hours or units, and whether the cumulative value still fits inside the order. Confirmation that the work was delivered goes to the person who ordered it, and invoices with no order at all are collected on a separate list with the amount and the cost centre.
The automation reads statements delivered through the bank file exchange and, following the agreed posting key, matches payments to invoices using document numbers, amounts and payment descriptions. Matched items are marked as settled in the ERP; for the rest it produces a variance report with a suggested match where only part of the amount agrees.
The automation pulls the overdue list, splits it by ageing bracket and sends reminders in the tone assigned to each bracket, from a gentle note to a final demand. Letters at the last stage are prepared for approval rather than sent on their own. It records contact history on the customer, pauses sending where an agreed repayment schedule is running, and prepares a decision list for finance.
The automation produces balance statements as at a given date, sends them for confirmation, collects replies and records in a register which items are confirmed, disputed or unanswered. After a set period it sends reminders and tabulates discrepancies with the amount difference for accounting.
The automation pulls overdue payables from the ERP, groups them by budget owner and hold reason, then notifies the people responsible. It separately highlights items where the company is losing an early payment discount or risking interest.
The automation reads transaction statements, assigns operations to users and cost centres, classifies spending by category and reconciles it with submitted invoices and receipts. Employees with missing documents get personal reminders listing the specific transactions, and the complete set is prepared for posting.
The automation checks that a travel claim is complete, covering tickets, accommodation, per diems and exchange rates, compares items against the travel policy and flags limit breaches. It prepares data for posting and reminds employees with outstanding claims, listing the open trips.
The automation maps the chart of accounts against the tags required by the ledger structure and points out accounts without a tag and fixed assets with missing fields. It generates a test file, runs it through the validator and hands over a list of gaps. The check repeats monthly so that gaps do not pile up before the filing deadline.
The automation prepares data from source systems, logs in to reporting portals, fills in forms and downloads submission confirmations, which it files with the case. It keeps an obligations calendar, warns ahead of each deadline and marks the filing as closed once submitted.
The automation runs the repeatable close steps: reconciles control accounts against subledgers, checks that documents in the workflow are complete, drafts accrual calculations by rule for approval and publishes checklist status. Items that do not agree are listed with the difference and the system where it originated.
The automation takes the receipts issued by the payer application and the cyclical reports the institution provides, checks their status and files them by period and document type. It reconciles the set against the list of submitted declarations and reports gaps to HR with the declaration they relate to.
After an HR decision the automation creates accounts, raises access requests, generates a document pack from templates, assembles the personnel file and notifies the teams preparing the workstation, equipment and training. On departure it reverses the sequence and reports the closure of every checklist item.
The automation compares rosters, time registration and the data handed to payroll, spots discrepancies in overtime, allowances and absences and passes the list on for clarification. Once approved it loads the complete data set into the HR and payroll system before calculation; the calculation itself stays in the payroll system.
The automation checks the HR system for expiring occupational health certificates, fills the referral template with employee data and workplace hazard factors, then sends it to the employee and their manager. It follows up after a set period and escalates if the examination is still outstanding before expiry.
The automation reads reports from forms and e-mail, creates a register entry, notifies the safety officer and managers and tracks the deadlines for accident documentation. Monthly it compiles incidents by location, cause and consequence.
The automation maintains a register of training and licence validity, notifies the employee, manager and training team in advance, books places with external providers and files certificates. It also produces a list of people who cannot be assigned to a task without renewal.
The automation takes data on new medical certificates from the channel provided to the payer, creates the absence in the HR system, updates the roster and notifies the manager about the period. It flags cases needing a decision, such as sick leave overlapping planned holiday.
The automation identifies employees reaching auto-enrolment, prepares registrations, records opt-outs and generates monthly contribution files for the financial institution. After submission it collects confirmations and reconciles amounts against payroll.
The automation links invoices to new and existing assets, prepares bulk component additions from a spreadsheet and transfers assets under construction into the target register once the investment closes. Every run is confirmed by a control report listing the inventory numbers created.
Based on approved requests the automation carries out value increases and changes to depreciation rates, tax rates and cost centres, item by item with a log entry for each. The result is confirmed by a report comparing the state before and after.
The automation compares count sheets with the ERP register, identifies missing items, surpluses and assets in the wrong location, then prepares discrepancy protocols for approval. It also tracks the stocktake schedule across locations.
The automation tracks inspection, insurance and testing dates, reminds vehicle users and files the certificates. In parallel it logs in to fuel supplier portals, downloads transaction statements, assigns them to vehicles and cost centres and prepares data for posting.
The automation collects requisitions from forms and e-mail, enriches them with budget data and the price history of the same item, routes them for approval by value thresholds and, once approved, creates the order in the ERP and sends it to the supplier. It chases missing delivery date confirmations.
The automation follows published industry indices, commodity and energy quotations and freight rates, and checks availability and stated lead times for key items in the supplier portals the company uses. It sets this against its own costs, stock levels and the level assumed in the budget, calculates the deviation and shows what changed since the previous run. Purchasing gets a signal when a threshold is crossed, and planning a list of items whose lead time has stretched enough that the order has to go earlier.
The automation builds the list of items expected in a period from contracts, orders and the settlement calendar, then compares it with what has actually arrived: progress reports, timesheets, acceptance protocols, statements and insurance policies. A reminder goes only to the employee or subcontractor with a gap, naming the item and the deadline, and after another run with no response the case goes to the contract owner. Gaps that payment depends on are listed separately before the payment batch is released.
The automation reads order confirmations arriving by e-mail and from supplier portals, compares them with the order in the ERP and flags differences in price, quantity and delivery date. Matching confirmations are recorded against the order, differing ones go to the buyer with the line marked, and orders left unconfirmed are chased with the supplier after a set time.
The automation pulls shipments without proof of delivery from the ERP, logs in to carrier portals, searches by dispatch document number and downloads the confirmation files. They are filed in a structure linked to the sales document, and shipments still unconfirmed after a set period are raised as claims.
The automation links the goods issue note with the consignment number registered at the carrier, records it on the sales document and, once delivered, adds the proof of receipt. Shipments with an incomplete set are listed with the reason: a goods issue never dispatched, a parcel sent with no goods issue, or no proof of delivery after the declared transit time.
The automation takes planned shipments from the ERP and notifies customers with the date, carrier and waybill number, choosing template and language to match the recipient. Send confirmations are filed with the document and invalid addresses are reported for correction in the master data.
The automation checks shipment statuses with carriers, spots holds, returns and breaches of the declared delivery time and raises claims with the full document set. It keeps a case register and tracks the carrier's response deadlines.
The automation splits the carrier invoice into lines, matches them to individual shipments by consignment number and compares the rate with the agreed tariff, the declared weight and the pallet count. Matching lines are prepared for posting, while fuel surcharges, weight corrections and extra fees with no basis in the tariff are compiled into a claim to the carrier.
The automation creates transport declarations from dispatch documents, adds vehicle registration and carrier details, retrieves reference keys and passes them to drivers and recipients. It monitors open declarations and prompts for completion before the deadline.
The automation pulls cost and stock data from source systems, compares it with plan and norms, calculates variances and prepares a report for controlling and logistics. It highlights items needing a decision: stock ageing beyond a threshold, shortfalls below minimum and budget overruns.
The automation logs in to customer booking portals, reserves time slots in line with the dispatch plan, enters vehicle and pallet details and downloads booking confirmations. Slot changes made by the customer are reflected in the plan and passed to transport.
The automation monitors the order mailbox, reads documents in PDF, spreadsheet or message body and recognises item codes, quantities and delivery terms. It checks completeness and price list consistency, prepares an import file and returns incomplete orders to the salesperson with the item in question marked.
The automation runs reports in the ERP and analytics system, downloads the files, merges them with spreadsheet and customer portal data, then distributes the results to sales and management in the agreed format. It checks data completeness and flags sources that were not refreshed.
The automation loads supplier price lists in various formats, recalculates them using agreed margins and exchange rates, compares them with current prices and prepares a change list for approval. Once approved it updates the ERP price list and publishes a version for the sales team, keeping change history.
The automation receives enquiries from the web form, e-mail and industry portals, creates a case in the sales system, matches the customer in the database and assigns an owner by region or industry. It confirms receipt to the sender and escalates cases with no response within the agreed time.
For a given list of companies the automation reaches into public sources: business registers, the financial statements repository, official gazettes and tender portals. It fills the record with legal form, activity codes, connected people and selected figures from the latest statements, and lists changes since the previous run separately. The data feeds the prospect list in the CRM and the trade credit assessment.
The automation logs in to key customers' supplier portals, downloads new orders, schedules and demand forecasts, converts them to the ERP format and loads them. Changes to existing orders are presented as a difference list for production planning.
Driven by HR events, the automation creates, modifies and disables accounts in the directory, mail, remote access and business applications, documenting every operation. It periodically reconciles active accounts against headcount and reports orphaned accounts.
The automation monitors data exchange queues, detects documents stuck in interfaces and safely retries processing where the cause is transient. Other cases go to the relevant team with the error description and document identifier, and recurring causes are summarised weekly.
The automation walks through key business scenarios in the test environment after each update, compares results with expectations and reports deviations with screenshots from the point of failure. The scenario set grows with each defect found, so every fixed bug stays guarded.
The automation checks expiry dates for TLS certificates, electronic signatures, domains and software licences, notifies owners in advance and raises a renewal task. It maintains a single obligations list with dates, cost and the person responsible.
The automation collects items from the electronic delivery mailbox, downloads proof of receipt, identifies the sender and case type, then creates a correspondence register entry and routes the document to the right team. It tracks response deadlines from the delivery date and warns as they approach.
The automation receives the request from a form or e-mail, opens a case with the deadline counted from the day it arrived and searches for the person across the systems listed in the record of processing activities: HR, sales, service desk and the document repository. It lists where the data appears, with the legal basis and the retention period, and for an erasure request prepares the items to close plus, separately, those that may not be deleted. The reply is drafted from a template and executed once the data protection lead approves.
The automation assembles monthly and quarterly reporting packs from prepared sources, formats them to the template, carries over the previous period's commentary as a reference point and distributes them to recipients. The set is archived with date and version.
The automation issues waste transfer notes from collection orders, confirms transports, updates records with actual weights from weighbridge tickets and ensures no note stays open. Ahead of the annual report it compiles the data in the required layout.
The automation reads fuel and energy consumption from invoices and registers, consolidates it by emission source, calculates reporting values and completes the forms in the reporting systems. It archives confirmations and keeps a deadline calendar with advance warnings.
The automation periodically checks key customers against insolvency and restructuring registers and official court and business notices, reporting changes to finance together with the current exposure. For new customers it runs the check before a trade credit limit is granted.
The catalogue does not cover everything. Describe your process and we will say whether it is a good candidate and where to start.
WHAT AUTOMATION CAN DO
Logging in, navigating records, filling forms, posting entries, searching by document number, exporting reports. Through the user interface or an integration, depending on what the system offers.
Reading messages by rule, saving attachments, analysing content, classifying and replying. Spreadsheets, PDFs with text recognition, documents from templates, data interchange formats.
Logging in to institutional, carrier, bank and customer portals, completing forms, downloading confirmations and reports. Where an institution publishes a programming interface, we use it instead of a browser window. Where a portal's terms rule out automation tools, the automation works on the files the portal provides and the login stays with a person.
Sending and retrieving data through programming interfaces, handling authorisation, database queries, comparing and validating data sets. Where an integration exists it is faster and more robust than clicking.
WHERE AI HELPS
Classic automation needs predictable input. When a process involves scans, free-form messages or documents in dozens of layouts, an AI model recognises the content, classifies it and extracts the fields needed, after which rules take over. This split keeps predictability where compliance matters and adds flexibility where data is irregular.
AI can be used in many of the processes in this catalogue, but it is not needed in every one. The form of the input data decides: where documents keep a fixed layout, rules alone are enough, cheaper to maintain and easier to explain during an audit. How far AI is involved is settled when the specific process is analysed, not upfront.
Decisions with financial or legal consequences stay with a person. AI prepares the material, it does not approve.
HOW WE WORK
We review candidate processes, estimate effort and point to the one that pays back fastest. You leave with a list ordered by return, not by who asked first.
We document the flow step by step with the process owner, agree the scope of exceptions and the access needed. The quote covers build, testing and support, with no items discovered midway.
The automation is built iteratively, with testing in a test environment after each stage. The process owner sees it running on their own data before anything reaches production.
Go-live with monitoring and alerts, then handling system changes and extending to further processes. Run history is available from day one.
THE TOOL
We build automations on our own platform, developed and supported in Poland. One environment covers the whole automation lifecycle: building the automation, schedules, task queues, monitoring and run history.
Orchestrator, automations and logs run inside your own infrastructure. For institutions and companies handling sensitive data this is usually non-negotiable.
Automations are Python scripts. The skills are available on the market, existing scripts can be adopted without rewriting, and you are not locked to one vendor.
Every execution leaves a record: what ran, when and with what result. The history is audit-ready without extra work.
Simpler processes are built in a generator, complex ones by the team in Python. Both paths run on one platform under one monitoring view.
WHAT WE OFFER
We take the process from analysis through to support and build the automations on the Testto Automate platform. Your team accepts the result on its own data and reports changes.
We deliver Testto Automate into your infrastructure, onboard the team and hand over the working standards. Further automations are then built in house, with our support for the harder processes.
In both cases the solution runs on-premise: orchestrator, automations, logs and data stay with you.
A free consultation takes about an hour. We review candidate processes, estimate the effort and point to the one worth starting with. No sales deck, no commitment.
The process catalogue and the qualification test are available on this page, with no form and no sign-up. It helps to bring one process to the call, the one that raises the most doubt.