We build, test and ship automation and AI that work in production.
Three services from one team: process automation and company data research that take manual work off your team, test automation for web apps and APIs, and AI built on your own data. From the first call to a running, tested system.
Three services that work together. Start with a small pilot on one problem, or let us take a whole system from idea to production. The price is fixed and agreed before we start.
01
Process and data automation
We take the copy-paste, retyping and manual checks out of your team's week. Data moves between your systems on its own, and a person approves anything that matters.
What you get
Workflows between your tools, APIs and databases
Company data from public registers: codes, sectors, headcount and decision-maker roles, cleaned, deduplicated and ready for your CRM
Daily monitoring of news feeds, tenders and public registers
Scheduled jobs, reports and a short morning digest
Hire us for one step or for the whole way. Testing is our day job, so every system we build comes with its own tests and CI pipeline.
01
Scope
A 30-minute call, then a written scope: one problem, what done looks like, and a fixed price.
02
Build
Automation, AI or a test suite, in your repositories from day one.
03
Test
UI, API, data and AI checks, written with the code, not after it.
04
Deploy
Docker and CI/CD, so every change is checked and shipped the same way.
05
Run
Monitoring, a handover to your team, and monthly support if you want it.
Work
Work we can show
Across all three services. Client names stay private, and you can ask about any of these on the call.
01Test automation
Release checks for banking, trading and e-commerce platforms
Several environments and complex payment flows, where one regression or flaky test holds up a release.
A Playwright and TypeScript framework used by several product teams. Selectors survive UI changes, suites run in parallel across environments in CI, and a failing check stops the release.
Playwright · TypeScript · API tests · CI / CD
For: Banking · Trading · E-commerce
Test automation · example project, real run
Example suite: 20 UI and API checks against an example shop with three planted bugs. All three were caught on desktop, mobile and API.One failure up close: the cart total ignores quantity. The report shows the expected value, the actual value and the exact line.
02Automation
Sales leads from Lithuanian news and public registers
Finding growing companies by reading news sites and registers by hand takes hours every morning.
10 Lithuanian news sources (VŽ, Delfi, 15min, LRT, Lrytas, Google News and more), deduplication, AI filtering, and one morning email with Registrų centras, Sodra and public procurement data for each lead.
From project 02 · 219 articles: rules first, then AI
One real run of our news filter (15 Jan 2026), headlines translated from Lithuanian, with the reason each one was dropped or kept.
03AI engineering
Polish business news, filtered by rules and AI
Hundreds of Polish business articles a day, and only a handful are real sales opportunities.
11 Polish business news feeds (Puls Biznesu, Bankier, Money.pl, WNP and more). A regex pre-filter drops most of the noise, then Claude Haiku and Sonnet classify the rest, with an Excel log of every decision.
Off-the-shelf voice bots speak English first and know nothing about the caller.
Lithuanian voice agents whose prompts are filled with live CRM data before every call, plus a React dashboard to run agents for several clients.
ElevenLabs · React · Node.js · PostgreSQL
For: Voice AI · Lithuanian
05Automation
Company data from public registers and email deliverability
Outbound email to unchecked lists bounces, lands in spam and damages the sending domain.
Company data from KRS, CEIDG and Rekvizitai.lt, merged and deduplicated, mailbox verification, SPF / DKIM / DMARC set-up, and Polish email sequences for several brands.
Python · Public registers · SMTP checks · DNS authentication
For: Outbound and data · Poland
06AI engineering
Flowly, an AI test generator for Playwright (prototype)
AI-written tests look right, then fail on the first run because the selectors are made up.
Plan first, then code: the tool maps the user flow, checks every selector against the real page, and only then writes the Playwright test.
LLM pipeline · Playwright · Selector validation
For: Own prototype
How we work
Four ways to start
Each one has a fixed scope and price agreed before we start. You own the code and pay no platform or licence fees.
01
CI test pipeline setup
Your tests running in CI on every push. Your existing tests, or up to 10 smoke tests we write, in GitHub Actions or GitLab CI, with a report and screenshots on failure, a short README and one handover call. Fixed price.
02
Pilot (1–3 weeks)
Your most painful problem: one workflow automated, one AI pipeline measured with evals, or one critical user journey covered by tests. Fixed scope and price.
03
Project
A full automation system, AI feature or test framework, from scope to production, with a handover to your team. Scope and price agreed after a discovery session.
04
Monthly support
Agreed hours per month for test maintenance, new automations and AI improvements.
About us
Engineers who test for a living
Day to day we build UI, API and database test frameworks for banking and other regulated systems, where a flaky test or a wrong number is expensive. We bring the same discipline to automation and AI: small, tested, measurable systems that your team owns.
You own the code
Repositories, pipelines and prompts stay in your accounts. No fees for a closed platform.
A person has the final say
Automation prepares and checks. Anything that reaches your customers or core systems is approved by your people.
Proof before pitch
Real test runs, real data logs and measured accuracy, not slides about what AI could do.
You talk to the people doing the work
No account managers. The people on the call are the people writing your code.
FAQ
What teams ask before the call
What can we hire you for?+
Process and data automation, test automation (UI, API, database) and AI engineering. Hire us for a 1–3 week pilot on one problem, or for the whole cycle from scope to production.
Can you work with our existing tests?+
Yes, that is the most common starting point. We review what you have, remove tests that add nothing, fix the flaky ones and wire the rest into CI.
How do you know an AI feature actually works?+
We measure it. Before launch we test the pipeline against real examples with known answers, and the same checks run after every change. Cheap rules handle what they can, so the model only sees what needs it.
How long does a pilot take?+
One to three weeks. We pick one problem, agree a fixed price, and deliver a working result in your environment.
Do you need access to production?+
No. We work in your test environments and CI. We only touch production if you want us to deploy or monitor something there.
What happens after delivery?+
Code, configuration and documentation stay in your repositories. We walk your team through it on a handover call. If you want, we stay on for monthly support.
Which languages do you work in?+
We work with teams across the EU in English and Lithuanian. The systems we build run in English, Lithuanian and Polish.
First step
In 30 minutes you will know if we fit and roughly what it costs.
Pick a time and tell us what is stuck: manual data work, flaky releases, untested APIs or an AI idea that never reached production. No commitment.
