Glitni | Summary of the First Half of 2026

We entered 2026 fully booked until the end of February. That sounds safe, until you think about what happens on 1 March. This is the story of the half-year where the countdown turned into a refill: what we achieved, and what we didn't get round to. As honestly as we can, in retro format.

“Fully booked” sounds like a state of being. In consulting, it’s a countdown to the moment we’re not. When we entered January, everyone was on assignment until 1 March. After that, the calendar was in theory gradually wide open for 4-5 of us.

Then March came, and with it came enquiries from insurance, banking, energy and grocery retail. Some found us through our articles, some through LinkedIn, some through people we’ve worked with before. The countdown was cancelled, and the half-year instead became one of the busiest we’ve had. So busy that we even brought in a few friends, whom we found assignments for.

Last year’s figures are publicly available at the Norwegian company register: NOK 15.7 million in revenue in 2023, 21 million in 2024 and 37.8 million in 2025. How 2026 ends, we’ll tell you once the figures are filed in the same place.

That growth means we qualify as a DN Gazelle this year. The Oslo list is published in November, and that’s a proud moment. What pleases us most is that we’ve grown this fast without diluting the professional level.

Highlights

• We’ve grown. From August we’re 16 at Glitni, and numbers 17 and 18 have signed. We’re ready to recruit more seniors when the right people become available.
• We’ve had plenty of assignments throughout the half-year, across data platforms, data engineering, data governance and training.
• We released our own data ingestion framework as open source in May, and shipped 29 versions of it over the half-year.
• We organised Oslo dbt Meetup number 11.
• The social calendar delivered from the first month: fire pit, relay race, concert and summer party.

Sales and projects

Sales and projects | Roles we’ve taken on at our clients

• data platform developers and architects
• data engineers and team leads for data engineers
• dbt experts
• advisers and project managers in data management and data governance
• instructors for upskilling in data and AI

Sales and projects | What we achieved

• Demand came to us: We signed new clients and expanded existing assignments, in energy, telecoms, media, banking and health technology among others. Most of the conversations started with the client getting in touch. The sales and marketing machine works, and that’s the finest confirmation our professional strategy can get. Thank you for the trust!
• Larger teams at large clients: We’re working our way in with several consultants at a time at several of our large clients. That gives better deliveries, more learning across the team, and colleagues who work together day to day.
• We delivered across the full breadth: Developers and architects on data platforms, advisers on data governance, and upskilling in modern data tooling for a large client. We also prepared a new round of a course programme we deliver together with a partner.
• We invested in our own tool: We put real work into our data ingestion framework, and released it as open source in May. Its own section below.

Sales and projects | Where we can improve

• We don’t share enough of what we learn about AI-assisted data engineering: some of us have become very good at using the new tools, but each on their own part of the workflow, and what’s possible varies with the client and their setup. We haven’t managed to turn the best practice into something everyone can pick up, meaning ready-made workflows, skills and agents rather than each of us reinventing it. That’s the job we’ll put the most into going forward.

The framework we gave away

The most concrete thing we did this half-year was to give something away.

A data engineering team spends a disproportionate amount of time on work that is the same from company to company. The code that pulls data from a source and lands it in a database has the same building blocks whether you sit in a media house, a bank or an online shop. It has to handle pagination, retry when something fails, fetch only what’s new, tolerate the source changing a field, and log what it does. Every team solves it from scratch, and a data engineer can spend weeks on code that should never have been written by hand.

dlt-saga is our answer to that. It’s a framework for loading and historising data, built on the open library dlt, also known as data load tool. The core idea is that you set up the common data sources with a configuration file rather than Python. Where we used to spend days connecting a new source, we now spend hours, and often less. The framework grew out of a client assignment, was proven in production there, and runs daily at that client today. We’re now rolling it out at client number two. In April we separated the framework from the client’s own pipelines, so the code could stand on its own. On 8 May we published the first version on PyPI under Apache 2.0. From then on, anyone could install it, read it and use it, without asking us for permission.

Why give it away? Because code is easier to believe in than a reference list. A client considering us can read the code and judge us by it. A data engineer considering joining us can see what we build. We charge no licence fee for the framework, and we expect to earn more from letting the framework become known than we would have earned from charging for the code.

Then things moved fast. Through May and June we shipped 29 versions. The framework gained its own command-line interface, history on snapshot tables, documentation and classification written all the way down into the database, and a report showing how the pipelines are running. Sindre Grindheim, assisted by AI, has written nearly all of it. More of us will contribute going forward. And as with every good open initiative like this, the whole world can now contribute.

The framework | Where we can improve

• We’re time optimists: the plan was a coordinated launch week with an article on why we’re giving this away, LinkedIn posts the same day, and a round in the dlt and dbt communities. The code has been open since May, and glitni.no still doesn’t have a single word about it. The official launch will happen, it has just taken longer than we set aside. We finish building before we get round to telling anyone.

Recruitment

Recruitment | What we achieved

• We reached 16, and numbers 17 and 18 have signed: Garth joined us in March, and on 3 August David started: that made us 16. Numbers 17 and 18 signed in August and September. We’re still a whole team contributing to recruitment, and that pays off over time.
• Many good candidate conversations: We spoke with a great many talented people this half-year. Some need a few more years under their belt before they’re the architects and senior developers we’re looking for, and we hope to meet them again. Quality comes before pace, even though we’d gladly have grown faster.

Recruitment | Where we can improve

• We need to keep the candidate pool warm at all times: Recruitment took longer than we’d hoped for certain roles, quite simply because we hold on to quality rather than hiring fast. A senior we wanted chose a leadership role elsewhere. The lesson is that candidate dialogues have to stay alive all the time, not just when a concrete need arises.
• We still need to work on the balance: As last year, some simple ratios about the team: 14 men and 2 women. Not good enough, in other words. Our ambition stands: at least 1 in 3 new hires should be women, and nearly everyone with us is a highly experienced senior.

Internal life

Internal life | What we achieved

• The social calendar ran itself: Fire pit at Grefsenkollen in March, the Holmenkollen Relay in May, the Over Oslo festival in June, group training sessions in between and a summer party at Fiskeriet just before the holidays. No event agencies, just people who want to get together. That’s perhaps the best sign that the culture is alive.

• Growth we can report after all: The 2026 figures have to wait, but we can say something about growth: several of us had their first child this half-year. A baby boom at Glitni, and we send our warmest congratulations.

• The Holmenkollen Relay deserves its own point, again: Last year we promised that we wouldn’t win this year either. We’re proud to report that we kept our promise. We made it to the finish line, the mood was good, and that’s plenty.

Internal life | Where we can improve

• The development reviews slipped: They should have been held with everyone before the summer, but they slipped, and they’re at the top of the list going into the autumn. When we grow, what concerns our people has to come before most other things.

Expertise and visibility

Expertise and visibility | What we achieved

• Oslo dbt Meetup number 11 took place in June, with good attendance and strong professional contributions, including from our own Anton. The meetup has become a regular gathering point for the data community in Oslo, and we’re proud to keep running it.
• The podcast Datautforskerne (Data Explorers, in Norwegian) continued with new episodes through the half-year. Not that many, but something is better than nothing.
• The articles on glitni.no kept up the pace until April on the topics we want to be known for: data platforms, data engineering and data governance.

Expertise and visibility | Where we can improve

• The podcast and the website still have plenty of plans that haven’t been carried out: We have more on the list than we managed to publish, and a whole pile of nearly finished material was left lying. We’re taking that with us into the autumn.

The road into the second half

The biggest question we’re working on right now is how to use agentic AI as a company. Not person by person, assignment by assignment, but as a shared way of working: shared tools, shared standards and shared learning. At the same time, we see the lines between disciplines blurring. Everything we do becomes “data and AI”, and that shapes both our services and how we develop our people.

Concretely, that means language models integrated into data engineering work: in how we work, and in what pipelines and data products end up looking like. That makes architectural understanding and the big-picture view more important, not less. We’re also seeing growing demand around AI governance: ambitions out there are high, and then the controls have to be in place. We’re working on making that a clearer service this autumn. If you’d like to hear how we think about this, get in touch.

On the visibility side, there’s more coming from us this autumn. We’re refreshing glitni.no, making more video content, and launching our own newsletter where we share what we learn about data and AI. And we’ll do the dlt-saga launch we owed ourselves in May. Follow us on LinkedIn and you won’t miss it.

Otherwise, we go into the autumn with a bigger team and good momentum in the field. The professional day in August and the mystery trip the first weekend of September are behind us: the whole team went to a secret location in Europe. And we’ll keep sharing knowledge through the podcast, articles and client cases.

Thank you to all clients, partners and colleagues for a good half-year. And to those of you considering getting in touch: the calendar is, in theory, blank from a date we’re not disclosing here.

Rather watch than read? Here is the half-year summed up as a video (in Norwegian).

Follow our journey

Here you can read similar articles:

Curious?

If this piqued your curiosity, please don’t hesitate to get in touch with us or follow us on LinkedIn!