Automate invoice processing, file review, Wwft checks, and standard reporting. More space for advisory work, less manual entry. Integrates with Twinfield, Exact, AFAS, Visma, and Yuki.
Accountancy firms and bookkeepers are under pressure. Demand for advisory work grows, rules get more complex (Wwft, CSRD, box 3, tax changes returning almost every year), and the inflow of new staff lags. At the same time there is still a lot of manual work in the process: coding receipts, preparing VAT returns, typing annual account explanations, sending standard letters for missing documents. That work has to be done, but it does not all have to be done by humans anymore.
AI changes that without changing the chain of responsibility. An incoming invoice receives a booking suggestion based on hundreds of previous bookings for the same client. A UBO check automatically runs against sanctions lists and adverse media and produces a first risk assessment. A draft letter about missing documents is ready before the assistant starts on it. A compilation file receives a first review with attention to known risk areas. The accountant assesses, authorises, and advises. AI does the preparatory work.
DataDream works for SME accountancy firms, administration offices, and compilers in Zeeland and the rest of the Netherlands. This is a regulated profession. NBA rules, Wwft obligations, AFM supervision for statutory audits, and the fundamental principles (integrity, objectivity, professional competence, confidentiality, professional behaviour) are hard constraints. Builds therefore use EU-only providers, with an audit trail supervisors accept, with retrieval on validated sources, and with a human in the loop on everything going to the tax authority, the client, or a supervisor.
Starting can be small. First a short scan is done to see where time leaks in your firm. Often it is not in the complex cases (where advisors actually add value) but in the hundreds of small tasks around them: chasing missing documents, processing mutations, answering simple questions between quarters. Upfront it is calculated what it returns in time per week and builds only happen where it makes a difference. No AI because it is hip, AI because staff get back to the work they were trained for. For in-house finance teams at SMBs (CFOs, controllers, and FP&A), see AI for finance teams.
Organizations working with DataDream
Despite scan-and-recognise in Twinfield, Exact, or Yuki, a lot of manual coding remains. Especially for SME clients with non-standard suppliers, split bookings, and VAT categories that are just outside default.
AI learns from previous bookings per client, suggests general ledger account and VAT category, recognises new suppliers, and flags deviations. The staff member authorises instead of typing, with an audit trail showing where the suggestion came from.
Explaining differences between quarters, checking whether all invoices are included, cross-checking VAT categories, preparing corrections: predictable work, but every quarter again.
AI compares returns across periods, flags deviations that need investigating, drafts an explanation, and marks files the compiler really has to look at. Not for sending, but for the preparation step.
UBO verification, sanctions lists, adverse media, periodic re-assessment, file building: an hour or more per client, and at volume it adds up. Yet it has to be done and documented for supervision.
AI collects and summarises the external checks, drafts a risk assessment with source references, and keeps an audit trail. The compliance officer or partner reviews and signs off. Time per file down, quality and consistency up.
Deadline reminders, document requests, requests for missing documents, year-end instructions: dozens of emails per client per year. Low-value work that pulls time from billable hours.
AI generates correspondence based on client status and firm templates, in your firm style, ready for a quick check. What took fifteen minutes per email now takes two minutes plus sign-off from the handler.
What is our position on that specific tax construction? Which case law have we used before? What does the latest NBA guidance say? The information is there, but spread across mailboxes, files, and handbooks.
DataDream builds an internal knowledge base where AI searches validated sources: own positions, NBA publications, tax handbooks, case law. Answers come with direct source references so the staff member can click through and verify.
Laurens helped us bring a programme offering to life that was yet to be launched. We worked very well together on this project. Here's to more great cases where we can use AI to improve our services!
Jordi Dooge
Business Scout, Dockwize
Automating accountancy with AI starts with the repetitive work: invoice processing with automatic booking suggestions, VAT preparation, standard correspondence, and file review. AI works out the first draft, a staff member checks and authorises. It connects to your existing package (Twinfield, Exact, AFAS, Yuki, Visma), so automating your bookkeeping does not mean switching systems. You start with one process that costs a lot of manual work and expand where it measurably saves time.
Yes. DataDream works with Twinfield, Exact Online, AFAS, Visma, and Yuki. Most packages have an open API or a supported connection, so invoice recognition, automatic booking suggestions, and reporting export can be plugged in without staff leaving their work environment. For packages without an open API an intermediate solution with scan-and-recognise or email-to-booking is explored. Builds only happen once the connection runs stably and the bookings are correct, otherwise it produces net extra work instead of less.
You remain professionally and legally responsible, that does not change. AI is a tool, not an accountant. A review step is therefore always built in: AI drafts the suggestion, a staff member checks and authorises. For VAT returns, annual accounts, and tax positions, a human is always in the loop before anything goes out. What AI does do: work out the first 80 percent so the senior or compiler does not start from a blank screen. That saves time without changing the chain of responsibility.
Client data is sensitive: financial figures, personal data, sometimes strategic information. DataDream works with EU-only AI providers where data is not used for training and not stored after processing. For firms with larger clients or strict DPA requirements a setup is built where data does not leave the firm network. Per use case it is documented which data is processed, by which model, and for how long. That is also useful for your own GDPR accountability and for questions from clients who want to know what happens with their figures.
Yes, and this is one of the places where it makes an immediate difference. AI can search UBO checks, sanctions lists, and adverse media, draft a first risk assessment, and record the outcomes in a file. The compliance officer or partner reviews the assessment, but the gathering and summarising work is done. Important: the final risk assessment remains human work, especially in elevated or unclear risk cases. The audit trail is tightened here so you can demonstrate to BFT or the tax authority during supervision which steps were taken and who reviewed what.
OCR reads text from a receipt. That is where it stops. AI can suggest, based on previous bookings, which general ledger account an invoice belongs to, recognise whether a supplier is new, flag a deviation between the receipt and what normally comes from that supplier, and generate descriptions consistent with your categorisation. The difference is in the prediction of the booking, not just in reading text. For an SME portfolio with many small clients this is the difference between "scan and type" and "check and approve".
Honestly: a general language model does not know what changed yesterday in tax law or in an NBA guideline. This is solved by connecting AI tools to validated sources, for example an internal knowledge base with current NBA publications, tax handbooks, case law via rechtspraak.nl, and internal firm positions. The AI then cites only from those sources, with direct references. For topics where rules move quickly (think box 3, business succession, CSRD implementation) the recommendation is to always let a human do a final check, even if the AI seems to give a correct answer.
Engagements run in phases. First a short scan to see where within your firm the most time goes into manual work, often invoice processing, standard correspondence, or file review. Then a focused pilot on one process, with a team that really uses it. Only when the pilot works and staff value it does the scale expand. You evaluate at each phase whether to continue. This prevents a tool being rolled out that no one uses, and lets the firm gradually adapt to new ways of working.
Book a free call or send us a message. We always respond within 24 hours on business days.
Phone
+31 85 124 95 22Location
Middelburg, Zeeland
Availability
Reachable 24/7 digitally
