AI integration · Estonia & worldwide

AI integration that takes routine work off your team

We connect AI to the documents, inboxes and systems you already use (invoices, orders, contracts, customer requests) so the repetitive part runs on its own and people only check what needs a decision.

  • Scope agreed before work starts
  • Human approves every posting
  • GDPR & EU AI Act aware

01

What we automate

Six services, one goal: fewer hours on work that follows the same pattern every time.

02

Large companies already use AI. Most small ones don’t yet.

In 2025 more than half of Estonia’s large enterprises used AI, against roughly one in five of those with fewer than 50 staff. The tools are the same for both; what small teams lack is time to connect them to their own processes. That part is our job.

Share of enterprises using AI, 2025
  • EU · small (10-49 staff)17%
  • Estonia · under 50 staff20%
  • EU · all enterprises20%
  • Estonia · all enterprises22%
  • Estonia · 250+ staff53%
  • EU · 250+ staff55%

Sources: Eurostat (isoc_eb_ai), Statistics Estonia ICT survey 2025. Enterprises with 10 or more employees. Estonia under 50 staff: “one fifth”.

03

Demo: 20 invoices, 26 seconds

We built a working invoice pipeline on 20 test documents designed to be awkward: four layouts, three languages, a phone photo, a crooked scan, an EU reverse-charge invoice, a VAT mistake, a duplicate and an advert. AI reads them, code checks every sum, and the accountant approves before anything reaches Merit.

17 / 20
invoices ready to post without corrections
3 / 3
problems caught: wrong VAT, a duplicate, an advert
200 / 200
fields read correctly (Claude Haiku)
26 s
for all 20 invoices, processed in parallel
Time per invoice, seconds
  • By hand · new supplier2 min 30 s
  • By hand · known supplier1 min 40 s
  • AI reading + checks≈1.3 s

Manual: our own timing of entering an invoice into Merit, including finding the file. AI: 20 demo invoices processed in 26 s. A person still reviews flagged documents.

Read how the demo works →

04

How a project runs

  1. First conversation

    You describe the process; we tell you honestly whether AI fits it and what it would take.

  2. Process audit

    We look at real examples, time the manual work and write down what to automate, the expected savings and the risks.

  3. Pilot on your data

    Usually 2-4 weeks. A working version on your own documents and systems, with accuracy measured against what your team would have done.

  4. Launch & support

    We move the pilot into daily use, train the people who review its output, and monitor accuracy and cost every month.

05

Where it fits

Finance & accounting

Purchase invoices, receipts, bank statement matching, expense reports, payment reminders.

Sales & CRM

Lead qualification, enriching contacts, drafting quotes and follow-ups, meeting notes into the CRM.

Customer support

Website chatbot, ticket triage, suggested replies, multilingual answers from your knowledge base.

Operations & logistics

Orders from email and PDF into the ERP, delivery notes, CMR and customs papers, stock queries.

HR & admin

CV screening support, onboarding questions, contract and policy search, timesheet checks.

Management & reporting

Weekly summaries from several systems, anomaly alerts, plain-language answers about your own data.

AI for accounting firms → · See example scenarios for other industries →

06

Built to be trusted with real work

  • A person decides

    AI prepares, people approve. Nothing is posted, sent or paid without a rule you agreed to or a human click. Mistakes in bookkeeping or customer messages are expensive to undo.

  • Your data stays yours

    We use provider APIs whose business terms exclude training on your data, send only the fields a task needs, and can keep processing in the EU or on your own server.

  • EU AI Act ready

    Since 2 August 2026 chatbots must tell people they are talking to AI. We build that disclosure in, keep logs of what the system did, and document how it is supervised.

  • No lock-in

    Code, prompts and accounts are yours. Models are swappable, so you can move to a cheaper or better one without starting over.

07

Public funding can cover part of the cost

The Estonian Business and Innovation Agency (EIS) runs support schemes for adopting AI and automating business processes; the last AI adoption round offered up to €20,000 per company with 20% self-financing. Rounds open and close during the year. If one is open when we start, we write the technical description and use case your application needs.

08

Questions

How much does AI automation cost?

It depends on the process, its volume and the systems involved, so there is no list price. After the first conversation we agree on the scope and give you the cost before any work starts.

How long until it works?

A focused pilot on one process typically takes 2-4 weeks, including measuring accuracy on your own data. Wider rollouts are planned in stages so you see results early.

Will AI make mistakes?

Sometimes, which is why we design for it: code checks the arithmetic and business rules, uncertain cases are flagged, and a person approves anything that changes your books or reaches a customer.

Which systems can you connect to?

Anything with an API or a file export: accounting software such as Merit, Xero or Directo, CRMs, e-shops, Google Workspace and Microsoft 365, email, Telegram, databases and in-house apps.

Do we need to change our software?

No. We add AI around the tools you already use. Your team keeps working where it works today.

Do you work outside Estonia?

Yes. We work remotely with companies across the EU and beyond.

09

Let’s look at your process

Tell us which task eats your team’s time: what comes in, what has to happen to it, where it ends up. We reply within one business day with questions or a first estimate.

10

The company

Opentech OÜ is a software company registered in Estonia since 2018. We have built and run our own web products for years. The same engineering now goes into AI automation for other businesses.