2025-05-13 Podcast - Aldis Ērglis

Introduction

Welcome to AI Ranch. I’m Evalds Urtāns, and today I’m joined by Aldis Ērglis, Country Manager and Chairman of Emergn Latvia. With more than 20 years in software and leadership, Aldis now oversees a team of 350 specialists in Emergn’s largest global office. He’s earned the Microsoft MVP award in Data Platform for six consecutive years and built the Baltics’ biggest tech community, hosting over 100 events for 2,500+ members. In this episode we’ll discuss how his machine-learning lab and community work are shaping AI adoption in Latvia and beyond.

Current state of AI in enterprise

Just few days ago official goverment review found out that just 17% of goverment institutions use AI solutions and whoping 55% of institutions have no plans to use AI solutions at all. (https://lrvk.gov.lv/lv/revizijas/revizijas/noslegtas-revizijas/maksliga-intelekta-ieviesana-un-izmantosana-latvija)

CleanShot 2025-05-13 at 23.24.03@2x

  1. Is this your experience as well? Your feedback?

  2. Where do you see the biggest gap between perception and reality in AI today?

  3. What’s the biggest myth about AI in business you’ve seen?

  4. What’s a common misconception leaders have about AI adoption?

  5. How do you spot hype versus real impact in AI projects?

 

State Revenue Service (SRS)

Emergn Latvia led the development of a taxpayer rating system for the State Revenue Service of Latvia. This system informs businesses about their tax payment discipline and was delivered alongside a major data modernization (migration to SAP HANA) that enabled real-time data and the introduction of machine learning analytics into tax administration. As a result, SRS can now update data near-instantly and apply ML in its Taxpayer Rating System, while fraud detection lead times dropped from nearly a month to minutes with high-risk transactions flagged immediately.

  1. What is the SRS taxpayer rating system and how does it work?

  2. How long did it take to implement and train the system?

  3. What were the challenges of implementing the system?

  4. How accurate is the system?

  5. Can you share any unexpected stories or feedback from businesses about their taxpayer rating?

  6. What are the next AI projects in SRS?

Emergn’s Machine Learning Lab and other AI projects

The Latvia office operates an internal Machine Learning Lab headed by Aldis Ērglis. This lab creates, tests, and experiments with new AI/ML tools for a variety of client use cases.

  1. You recently ran a year-long GitHub Copilot trial with several Scrum teams. What measurable impact did Copilot have on productivity and code quality?

  2. Your lab built a prototype ML classifier for context-aware notification prioritization, aiming to cut productivity loss from pop-up interruptions. Using metadata (time, device, business object) and user feedback, it decides whether to show or delay each alert. What did your research reveal about the true cost of interruptions, how well did the prototype perform in practice, and what are the next steps for this technology?

  3. The Latvian Court Administration recently deployed a hybrid machine-teaching and NER pipeline to anonymize court judgments. The solution processes documents up to 20× faster—cutting average anonymization time from ~20 minutes to under 1 minute—and automates 80–90 % of cases, saving roughly 1,000 staff-days per year. Can you walk us through how this system works, what factors drove its success, and what lessons other organizations can take from it?

  4. BALTA Insurance has rolled out an intelligent email system built with a low-code Power Platform classifier. It now automatically processes 50 % of incoming vehicle-claim emails, reducing manual effort by 70 % and freeing the equivalent of 1.3 full-time employees per day for higher-value tasks. How was this project conceived, and what challenges did you face during implementation?

  5. Are there any other noteworthy AI projects underway at Emergn Latvia that you’d like to highlight?

Product mindset

A product mindset is the foundation of a strong organizational culture, this something you have been doing for a long time.

  1. What is the product mindset?

  2. What benefits have you seen from this approach?

  3. What practical steps can leaders take to make it a core part of their company's DNA?

  4. How do you instill this way of thinking across all departments?

  5. What are the most common mistakes leaders make when trying to implement a product mindset?

Leadership and community

Leadership is important not only within organizations, but also in the communities we are part of and help to shape.

  1. You're known for leading a large tech community in the Baltics. What's the most memorable or funny thing that's happened at one of your AI meetups or events?

  2. What's a leadership mistake you made early on that you can laugh about now, and what did it teach you?

  3. Do you have any daily habits or routines as a leader that others might find odd, but that you believe actually help you succeed?

Fun questions

Towards the end of podcast let’s do bunch of fun questions.

  1. Monday 9 AM — cruel or fantastic?

  2. What’s your go-to snack during late-night hackathons or product launches?

  3. Work-life balance: yes or no?

  4. What tech buzzword or jargon do you secretly wish you could banish from all meetings (at least for a day)?

  5. Which classic piece of career advice should be officially canceled in 2025?

  6. Metrics: do they usually guide or mislead?

  7. Data privacy or profit?

  8. Is AI innovation moving too fast or too slow?

  9. What’s one tech trend or innovation that still makes you feel as excited as a kid in a candy store, even after years in the industry?

  10. What’s the weirdest use of AI you’ve seen in the last year?

  11. If you could automate any daily chore in your life with AI, what would it be and why?

  12. If you could instantly download any skill into your brain Matrix-style, what would you pick?

  13. Finish the sentence: “The customer is always ___.” (No corporate Bullshit.)

  14. Which tech guru is overrated and totally needs an ‘unsubscribe’ button?

  15. Should the “Reply All” button be deleted?

  16. Would you use ChatGPT to write a birthday card?

  17. Have you ever quietly Googled or used ChatGPT to find a term during a board call—then pretended you knew it all along?

  18. Which emoji best describes your leadership style?

  19. What’s more terrifying: a critical production bug at 2 a.m. or karaoke night with a potential client?

  20. If you had to explain your job to a five-year-old, what would you say?

  21. If you could host a dream panel discussion for your AI community, which three people (alive, historical, or fictional) would you invite, and why?