VALA

22/10/2026

Vihreät testivalot eivät vielä kerro, toimiiko softa.

Kun tekoäly tuottaa koodia ja testejä nopeammin kuin kukaan ehtii arvioida, laatuammattilaisen rooli muuttuu, m***a se ei katoa.

Laadun takana -podcastin 20. jaksossa Niko Nousiainen (Hidden Trail) ja Virpi Tuohisto (VALA) puhuvat laatumittareista, testiautomaation ylläpidosta, ymmärrysvelasta ja siitä, miksi parhaista kehittäjistä ja testaajista ei kannata luopua.

🎧 Jakso on kuunneltavissa YouTubessa, Spotifyssa ja Apple Podcastsissa.

14/10/2026

Last Thursday we had our largest and most successful client event so far! Here are some vibes of it. Next one is already in the works; and as meeting fellow QA people is perhaps the greatest aspect of these, we're going to make it even bigger 🙌 !

09/10/2026

We couldn't have asked for more from our client event yesterday. Over 80 attendees, great speeches and most importantly our guests got to connect and hang around with like-minded people from other companies, some all the way into the night 😉! Next week we'll post our aftermovie but here are some photos already now.

02/10/2026

VALAn toimari Marcus Mattilan mielipidekirjoitus jälleen Tivissä. Toistamme VALAlla paljon tätä samaa mantraa, koska se on tärkeää. Alamme ihmisten on tärkeää ymmärtää, että QA alana on AI-ajan voittajia. Tulevaisuus on luonnollisesti epävarma, m***a me VALAlla uskomme, että testaamisen ja laadun osaajille on jatkossa paljon tarvetta.

Kuten Marcus tekstissään sanoo: "Koneelliseen laadunvarmistukseen siirtyminen ei vähennä asiantuntemuksen tarvetta. Se siirtää asiantuntemuksen sinne, missä sitä on vaikeampi automatisoida: laadun määrittelyyn, riskien priorisointiin ja siihen, että joku ymmärtää, mitä käyttäjä oikeasti tarvitsee. Ne yritykset, jotka rakentavat tämän kerroksen huolella, pärjäävät. Ne, jotka luulevat koneen hoitavan laadun ilman sitä, tuottavat vain nopeampia virheitä."

Linkki kirjoitukseen: https://www.tivi.fi/uutiset/a/6512bc60-025a-438a-98f2-121418a31633

--------------

In english, in short: Our CEO Marcus Mattila wrote another opinion piece in Tivi. It's only in Finnish but the point is that we need humans in QA work in the future too. This is an important theme for us and we keep talking about it because it's important that people in our field understand that QA is truly one of the winners in the AI revolution.

28/09/2026

Since LLM applications went into production, also experienced SW professionals have been asking how to test them. The old approach does not fit. When the same input can produce a different output every time, a pass or fail test tells you very little.

EvalOps, evaluation operations, is our answer. Instead of checking whether the output matches an expected string, you score it against defined criteria and thresholds, and you keep measuring in production, not only before release.

We are writing a blog series on how this works in practice. The third part covers output intelligence.

https://www.valagroup.com/blog/evalops-output-intelligence/

21/09/2026

Since LLM applications went into production, also experienced SW professionals have been asking how to test them. The old approach does not fit. When the same input can produce a different output every time, a pass or fail test tells you very little.

EvalOps, evaluation operations, is our answer. Instead of checking whether the output matches an expected string, you score it against defined criteria and thresholds, and you keep measuring in production, not only before release.

We are writing a blog series on how this works in practice. The second part covers protecting the AI model.

https://www.valagroup.com/blog/defensive-operations-and-live-validation-building-the-defenses/

16/09/2026

Nobody plans a QA career. Jamina Pöllänen got into tech through the sixth and last slot of her university application. Today she is a QA Lead and Tester of the Year.

Thanks Jamina for joining us 👏

16/09/2026

The traditional VALA midnight run didn’t disappoint: 10+ VALA runners, a few clients running with us, one Santa and five reindeer!

11/09/2026

Since LLM applications went into production, also experienced SW professionals have been asking how to test them. The old approach does not fit. When the same input can produce a different output every time, a pass or fail test tells you very little.

EvalOps, evaluation operations, is our answer. Instead of checking whether the output matches an expected string, you score it against defined criteria and thresholds, and you keep measuring in production, not only before release.

We are writing a blog series on how this works in practice. First part covers system robustness.

EvalOps blog 1. How to secure the deterministic backbone to enhance AI application quality. Six technical pillars, from prompt versioning to cost monitoring, essential for AI system robustness.

08/09/2026

Are you building test automation in a Salesforce environment?

Treating Salesforce like any other web app can make your test suite flaky, expensive, and difficult to maintain.

In this blog we share our experiences of what tends to break Salesforce test automation and what actually works.

https://www.valagroup.com/blog/why-salesforce-breaks-test-automation-and-what-works/