Process automation
As an expert-comptable (French chartered accountant), I personally build the automations that remove manual re-keying: Power BI reporting, cash flow statement, reconciliations, invoice collection, payment reminders. For finance teams as well as small businesses and the self-employed. You keep your tools, and you stay in control.
Finance teams
The same extracts, reworked by hand every month, then copied from one file to another: that is where time and reliability are lost. I automate these finance processes in your tools, so that your teams can focus on analysis.
Extracts from your accounting software or ERP feed straight into the reporting. A refresh replaces copying, pasting and reformatting.
The cash flow statement is rebuilt from the trial balances, using a mapping table between accounts and cash flows. Each update checks that the flows do explain the change in cash.
Balances and flows from each company come into a single tool, where they are classified by rules, consolidated and projected, with no re-keying.
Banks, debt, VAT, flows between group companies: the sources are compared line by line at each update, and differences stand out on their own.
Calculations redone at every close become models fed by the trial balance: same rules and same checks, from one financial year to the next.
The file that only one person knows how to run becomes a tool: clear structure, robust formulas, identified sources.
I know where the figures come from, and therefore where they are likely to be wrong. These checks run at each update and flag anything out of the ordinary, instead of leaving you to look for it.
Reconciliations and cross-checks bring differences to light: your accounting teams decide on the corrections. When a difference raises a tax question, I analyse it with you as part of an advisory assignment, also carried out by EZM CONSULTING. These automations can also be part of anoutsourced CFO assignment, full-time or part-time (time-share).
Small businesses and the self-employed
When you run your business alone or with a small team, repetitive admin eats into your core work. I automate what can be automated, with simple tools sized to your needs.
Your purchase invoices are retrieved automatically and matched to your bank transactions. All that is left to chase are the receipts that are genuinely missing.
Your bank is synchronised with your accounting software, Pennylane for example. Recurring transactions follow classification rules agreed with whoever keeps your books.
An accepted quote becomes an invoice with no re-keying, with the mandatory information, continuous numbering and tracking of payments received.
Late invoices are identified and chased on a schedule agreed with you, in a tone you approve.
Between businesses, late payment penalties can apply without a prior reminder, and a fixed €40 recovery fee is due for each invoice unpaid on its due date. Your reminders can point this out, in the tone you choose.
A photo of the receipt is enough: the expense is classified and tracked through to reimbursement, with no spreadsheet to copy out.
Revenue, cash received, cash position: a dashboard fed by your data, which you can check at any time.
Generative AI
AI saves time on text and documents. It replaces neither professional judgement nor checks: I use it where it brings a real gain, within a framework you approve.
My day-to-day work
I run a one-person firm, with no staff, and that is a choice: in my practice, automation does the work a team would do. The methods I set up for you are the ones I apply every day in my own firm. Since 2025, I have also been working within the finance department of a major property developer.
Examples
Four common situations, to give an idea of what changes day to day. The time that remains is time for analysis and decisions.
Before
About 2 days a month
Every month, the trial balances are extracted company by company, copied into the reporting file, adjusted by hand and then checked.
After
About half a day a month
The trial balances are dropped into a folder. Power Query combines them, applies the mapping table and runs the checks, then the dashboard updates. The time left goes into analysing variances.
Before
About 3 hours a week
Balances are noted bank by bank and copied into a spreadsheet, then upcoming flows are updated from emails and schedules.
After
About 1 hour a week
Statements are imported and classified by rules, and the 13-week forecast recalculates and flags low points. What remains is to approve the week’s assumptions.
Before
About 5 hours a month
Invoices are downloaded one by one from supplier portals and then sent to the accountant. Late-paying customers are chased case by case.
After
About an hour and a half a month
Invoices are collected and matched to bank transactions automatically, and reminders go out on an agreed schedule. What remains are the missing receipts and special cases.
Before
About 3 hours a week
The same questions arrive by email (appointment slots, documents to provide, terms) and each reply is written from scratch.
After
About 1 hour a week
AI prepares a draft from your instructions and your templates. You review it, adjust it if needed and send it yourself.
Illustrative examples, built to show orders of magnitude: they do not describe real assignments and are not a commitment to any result. The time saved depends on your volumes, your tools and the quality of your data. For completed assignments, anonymised, see the case studies.
My method
I start from your tools and your actual tasks. Each step produces something usable, and you keep control of the priorities.
With you, I list the tasks that come back every week, every month or at every close: who does them, how long they take, what errors they cause.
Each option is assessed in hours saved and in set-up effort. I start with what frees up the most time for the least effort, with controlled risk.
A first version runs on your real data. You use it alongside your current method and we compare the results.
Once the results match, the automation replaces the old method: access, scheduled updates, checks and alerts in place.
Each automation is documented: what it does, where the data comes from, what to do if it stops. The aim is for you to keep it running without depending on me.
After an initial conversation, you receive a written proposal: scope, deliverables and timetable.
Describe my processesTools
The choice of tool depends on what you or your teams will be able to keep running. I favour what is already in place in your business and what stays easy to follow for whoever takes over the tool.
The tool your teams already know: structured tables, formulas that refer to columns by name, assumptions kept separate.
The right choice when the deliverable must remain editable by your teams.
Imports, cleans and combines your extracts: trial balances, general ledgers, FEC files, bank statements. Each step is recorded and replayed at each update.
The right choice when the same extract is reworked every month.
Shared dashboards, refreshed automatically, with access rights depending on the reader.
The right choice when several people follow the same indicators.
Connects your applications: email, file storage, invoicing tool, spreadsheet. An event triggers a series of actions.
The right choice when a task consists of passing information from one tool to another.
Connected accounting: synchronised bank, invoices collected and matched, invoicing of your customers.
The right choice when a small business wants its accounts kept up to date continuously.
Drafting, document analysis, help with writing formulas and scripts, always reviewed.
The right choice when the work involves text or unstructured documents.
Claude Code, Anthropic’s programming assistant, used in the terminal: it helps me write, test and run custom scripts (processing large files, specific connections between tools, scheduled tasks), which I review before anything goes live.
The right choice when no standard tool covers the need, or the processing must run on its own at regular intervals.
Frequently asked questions
I ask only for the access that is strictly necessary, read-only wherever possible, and you can withdraw it at any time. Your data stays in your tools as far as possible: the automations run in your environment, not on a file passed around by email. Confidentiality is covered in the contract, and I can sign your non-disclosure agreement. When personal data is processed on your behalf, article 28 of the GDPR provides for a written contract setting out the processor’s obligations: processing on documented instructions, confidentiality, security, deletion or return of the data at the end of the assignment.
In most cases, no. I start from the tools you already use: Excel, your accounting software or ERP, your bank, your email. If a tool is genuinely missing, I tell you, with the possible options and their limits, and you decide.
It depends on the process and on access to the data. As a guide, a first targeted automation, for example preparing a monthly report or collecting invoices, is up and running within a few weeks: a review lasting a few days, a prototype used alongside your current method, then go-live. The precise timetable is set out in the written proposal.
Your teams, if you wish: each automation is documented (sources, rules, what to do if it stops) and I train the people who take it over. An automation can stop: a supplier changes its invoice format, an extract gains a column, an access expires. Checks flag the most likely anomalies, to limit the risk of a wrong result going unnoticed. Follow-up can also be included in the proposal, for fixes and changes.
Yes. When the business owner handles the admin, every hour saved counts. For a small business or a self-employed person, I start with invoice collection, bank synchronisation, quotes and invoices, payment reminders and a simple dashboard. The tools stay light and can be used without any particular technical skills. If your business needs closer financial management, outsourced CFO services on a time-share basis can take over.
Only with your agreement. Before any use, I tell you which AI tool I use and on what terms, and you decide. Confidential or sensitive information, such as health data, personnel files or full bank details, does not go through tools we do not control. AI prepares drafts, sorts documents, helps write a formula: it decides nothing. Its suggestions go through an approval step, by you or by the person you appoint. Amounts, for their part, are calculated by explicit rules that can be checked.
No. Automation assignments are carried out by EZM CONSULTING, whether your accounts are kept in-house, by your chartered accountant or by KOADVISE. If you already have a chartered accountant, I work with them: the tools put in place make their work easier without replacing them.
Yes. Since 1 September 2026, VAT-registered businesses established in France must be able to receive electronic invoices, through an approved platform. Large companies and mid-sized companies must also issue them from that date, and SMEs and micro-businesses will have to do so from 1 September 2027. Since your invoicing tools have to change anyway, it is a good time to connect quotes, invoices, payments received, reminders and accounting, so that each invoice is entered only once.
Another question? Write to me or read thefinancial, tax and accounting news (in French).
Further reading
Describe your processes and tools to me: you receive a written proposal, with no commitment.
