Recordly

✈️ 𝗦𝗽𝗲𝗮𝗸𝗲𝗿 𝗵𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁: 𝗙𝗶𝗻𝗻𝗮𝗶𝗿 𝗮𝘁 𝗗𝗮𝘁𝗮 𝗟𝗲𝗮𝗱𝗲𝗿𝘀’ 𝗕𝗿𝗲𝗮𝗸𝗳𝗮𝘀𝘁 How do you modernize data operations in a complex, fast-moving business like aviation? We’re excited to announce that Venkata Gowri Sai Rakesh Kumar Varanasi, Data Platform Architect and Markus Ekman, Business Analyst at Finnair will take the stage to show how they leverage dbt Mesh to remove data siloes and optimize airline operations. This is your final call to register for Data Leaders’ breakfast by dbt Labs and Recordly. Boarding starts on March 3 at 8.00 am. 🛫 Online check-in is open 👉https://lnkd.in/dgRGYjqY

𝗕𝗶-𝘄𝗲𝗲𝗸𝗹𝘆, 𝗼𝘂𝗿 𝗱𝗮𝘁𝗮 𝗮𝗻𝗱 𝗔𝗜 𝘁𝗲𝗮𝗺 𝗽𝗮𝘂𝘀𝗲𝘀 𝗳𝗼𝗿 𝗮 𝘀𝘁𝗿𝗼𝗻𝗴 𝗰𝘂𝗽 𝗼𝗳 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗮𝗹 𝗶𝗻𝘀𝗶𝗴𝗵𝘁𝘀 ☕ This is what kept them busy last week: 🟡 Tuomas Leidén took us deep into his 𝗰𝘂𝘀𝘁𝗼𝗺-𝗯𝘂𝗶𝗹𝘁 𝗳𝗶𝗻𝗮𝗻𝗰𝗶𝗮𝗹 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀 𝘄𝗲𝗯 𝗮𝗽𝗽 Stockastat, a product of curiosity, technical tenacity, and sense of real market need. What makes it special is:   • 𝗦𝘁𝗮𝗰𝗸: Django backend + Postgres for scalable and affordable ops   • 𝗙𝗲𝗮𝘁𝘂𝗿𝗲𝘀: Customizable metric/indicator creation, portfolio analysis, trend & peer comparison, and AI-powered stock narratives (with Gemini 2.5 Flash)   • 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴: A full-stack build means you live through the entire dev lifecycle, from user feature needs, to API integration, to admin dashboards and cost control. Respect for engineering discipline and for keeping feature creep in check! 🟡 𝗥𝗲𝗮𝗹-𝘄𝗼𝗿𝗹𝗱 𝗶𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻 𝗶𝘀 𝗺𝗲𝘀𝘀𝘆: Data ingestion is organizational as much as technical. Sometimes the “API” is an Excel file! Embrace small wins and aim for a shared, scalable process.   • Our hackathons drive ideas to production: Our recent hackathon led directly to improvements in real authority reporting and even kickstarted a computer vision engagement   • Embeddings & anomaly detection: Experimentation with embeddings for anomaly detection (e.g., in payments fraud systems). The lesson was there are always new tools to try, but don’t reinvent the wheel. Instead, start simple, experiment, then scale. 🟡 𝗠𝗼𝗱𝗲𝗹 𝘀𝗲𝗹𝗲𝗰𝘁𝗶𝗼𝗻 𝗺𝗮𝘁𝘁𝗲𝗿𝘀: “Don’t use a bazooka to kill a fly”. Powerful LLMs (Opus/Sonnet) are amazing, but the cost adds up fast. For routine tasks, Gemini 3 Flash gives great bang-for-buck. Our practical tips:   • Start with a plan using more capable (expensive) models if needed, then build out with lighter ones   • Share learnings about prompt/token efficiency (for ex avoid unnecessary tool descriptions in context, leverage skills in meaningful places over MCPs for integrations)   • Transparency in spend and keeping a “good enough” mentality is key to scaling responsibly. 🟡 𝗖𝗼𝗿𝘁𝗲𝘅 𝗖𝗼𝗱𝗲 - 𝗿𝗲𝗮𝗹-𝘄𝗼𝗿𝗹𝗱 𝗶𝗺𝗽𝗮𝗰𝘁:   • For Snowflake-centric workflows, Cortex Code streamlines query-writing, debugging, and data management. It’s a force multiplier, if identity management allows   • Security posture and access controls often trump raw innovation.  Communicating limits, documenting workarounds, and pushing for change at the org/process level can be solutions. #DataScience #DScoffee

AI has found its place in Master Data Management as a way to automate the parts of it that are slow, repetitive, and painfully dependent on individual experts, such as matching records, detecting quality issues, understanding complex ERP/CRM data models, and keeping documentation up to date. 🔍 Aino Vaittinen wrote an article that goes through practical ways AI can improve MDM and just as importantly, where it shouldn’t be used. ⚖️ If you’re working with MDM, governance, ERP/CRM integrations, or data platforms, we think you’ll find it useful. 👇 You can read it here: https://lnkd.in/dPTktPfP #BusinessDataCompany #MasterDataManagement

This week in our 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 𝗮𝗻𝗱 𝗔𝗜 𝗰𝗼𝗳𝗳𝗲𝗲, the team discussed simple yet impactful internal tools and streaming voice agents. 🤎 We took a look at our new, internally developed, billing agent that tracks project hours, creates billing drafts, and nudges team members when their hours are missing. For us, it was a good reminder that small, focused internal tools can have an outsized impact. Automated “nagging” frees people up to focus on their actual work. And reducing friction in everyday processes often delivers the fastest wins. ✌️ We also dug into recent developments in streaming voice agents. The bar is clearly rising; ChatGPT’s voice mode is setting expectations for natural, fluid conversations. At the same time, open-source alternatives like NVIDIA PersonaPlex are gaining attention. Direct speech-to-speech models enable real-time interaction with interruptions, back-and-forth, and those small “thinking aloud” moments that make conversations feel human. 🙋🏼 This shifts us away from speech-to-text → LLM → text-to-speech pipelines toward end-to-end, real-time conversational models. Rasmus Toivanen, our AI CTO, is known for his research on text-to-speech in Finnish language, so check his page out for more news like this. 🤖 𝗛𝗲𝗿𝗲 𝗮𝗿𝗲 𝗼𝘂𝗿 𝗽𝗿𝗮𝗰𝘁𝗶𝘁𝗶𝗼𝗻𝗲𝗿 𝘁𝗮𝗸𝗲𝗮𝘄𝗮𝘆𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗶𝘀 𝘄𝗲𝗲𝗸. ☕ Internal automation doesn’t need to be flashy to be valuable, as solving real pain points compounds quickly ☕ Streaming voice models are evolving fast, and expectations for “natural” interactions are changing with them ☕ Open-source voice and agent tooling is lowering the barrier to experimentation and custom use cases. Anyone here experimenting with streaming voice agents or speech-to-speech models? 🗣️ #DScoffee #DataScience #SpeechToSpeech

We just finished our 23rd hackathon in Tampere 🔥 In the picture you can see some new faces; five of our new-joiners got to experience their first Recordly hackathon today!
 We had a really impressive line-up of tracks this time. In fact, there were even tracks inside the tracks! Learning and trying out new stuff together is one of the greatest traditions we have at Recordly. 📚 Here’s some of the topics our team worked on: 🔹Demand forecasting & stock optimization: Worked on a PoC in which we aimed to maximize the business outcomes by making better stocking decisions 🔹Agentic authority reporting: Exploring how AI and agentic approaches can be applied to authority reporting when requirements are strictly defined by regulators 🔹Demo creation workshop with agentic approaches: One example was a real-time translation tool to solve the scarcity of interpreters needed for example at police interviews 🔹Agentic dbt & SQL development: Discovering a range of different cases, such as agentic migration workflows and ‘agentic data engineer’ 🔹The sales team took a look at what’s ahead in 2026 and focused on analyzing their choices, wins and losses. Moreover, we’ve enjoyed a wide range of local delicacies (not including mustamakkara this time) and next up we’re heading out for wings, darts and pool! 🎱 Essi the dog was also there to brighten up the day. Woof woof. We also like to share these moments with others; we’d love to arrange a tailored hackathon for YOUR team across multiple data & AI topics. Hit us up if you’d like your team to have this experience. 🤝

How do you get your data operations to deliver business value? If you don't know, then it's your lucky day because that’s the theme of our freshest ‘Set the Record Straight’ newsletter ⭐️ For this issue, we have some new blood in the editorial. Jenni Keskinen is a senior ML engineer and a new joiner at Recordly, and she'll give you her top 3 tips to succeed in data work this year. 🆕 Rauno Paukkeri, Co-founder and COO, opens up with his thoughts on impactful, value adding data work now and in the future. 🔮 And our CTO and data architect Mikko Sulonen gives his answers to some critical questions around agentic data development in an interview. 🕵️♂️ As always, our newsletter is available in Finnish and English. If you haven't already, subscribe! What would you like to read about in our next newsletter? 💌

Why did SOK decide to move to dbt and how did they manage it? Come to listen to Ville Antila, SOK’s Data Engineering Team Lead and find out: 🟢 How to navigate scope, complexity, and reality in a data platform project 🟢 The principles that helped the team get the migration done 🟢 Lessons learned: what worked, what didn't, and what surprised the team 🟢 Hands-on technical insights you can use 🟢 What the next phase looks like post-migration Sign up fast to secure your place in this breakfast event (link in comments) 💚

Big congratulations to Mika Heino for being recognized as a Snowflake Data Superhero! 🦸♂️ This global list highlights “data professionals whose creativity, deep technical expertise, and selfless community contributions are actively shaping the future of the AI Data Cloud.” 🐻❄️ This recognition reflects a lot of the hands-on work Mika has done over the years. Through his generous and open knowledge sharing, especially within the Finnish community, Mika has had a real impact, so we couldn’t be happier to see his name on the list of 126 Superheroes. Congrats as well to everyone included in the Class of 2026! ❄️ In true Finnish style, Mika himself commented he’s grateful for the recognition, and that he’ll be wearing his new jacket with style. At Recordly, we’re proud to have Mika as our Snowflake tech lead and to work closely with Snowflake as a Growth Partner. We empower our clients to harness the full potential of their data, ensuring they’re not just keeping up with their industry, but leading it! When you’re looking for a trusted Snowflake partner, hit us up 🤙 (If Snowflake’s Data Superheroes is new to you, it’s comparable to Microsoft’s Most Valuable Professional (MVP) program, which also recognizes community leaders who actively help others succeed with its technology.) See all professionals recognized in 2026: https://lnkd.in/gfunA5Eq 

How does MLOps actually work in Databricks? ⚙️ Our Databricks Lead and ML Engineer Jussi Virtanen shares a practical overview of DataOps, DevOps, and ModelOps on the Databricks platform. He walks you through each step and tells you what the necessary prerequisites are. 🧩 Enjoy 👉 https://lnkd.in/d2H6Z2ds #SetTheRecordStraight #BusinessDataCompany

At our upcoming event Data Leaders’ Breakfast on 3 March, our CTO Mikko Sulonen and Riccardo Di Mauro, Director Nordics, dbt Labs, will deliver a hands-on talk together for data leaders and practitioners who want to understand how modern data teams are evolving right now. 🎤 They’ll teach you how agentic workflows can help your team’s productivity and discuss the human focus shifting from execution and iteration to a deeper understanding of the business problem. 🧠 Come and learn: 🔶 What agentic workflows mean for data teams in practice 🔶 How productivity (and cost structures) change when agents becomes part of daily data development 🔶 How the human role shifts from execution to problem framing, validation, and business understanding 🔶 What this means for teams already using dbt or are considering moving to dbt. 👉 Don't miss this, save your seat now: https://lnkd.in/dgRGYjqY (You can also register via our LinkedIn event!)