Hannu Valtonen sold his last startup to Aiven, watched it become Finland's open-source darling, and then walked away to build something he won't quite call a competitor to GitHub. He says competitor sounds combative. Then he describes Avrea, his new company, in terms that sound a lot like, well, a competitor.

Avrea emerged from stealth on Tuesday with $4.7 million in pre-seed funding led by Earlybird. Reports out of Helsinki say the round closed in a few weeks, without a pitch deck, after multiple term sheets. That's the kind of detail founders love to drop into a press release. It's also, occasionally, true.

In this case there's enough credibility behind the founders to make the no-deck claim plausible. Valtonen co-founded Aiven, which scaled into a billion-dollar managed open-source data company. His co-founder Juha Valvanne was on the founding team at Nosto, the e-commerce personalization company. Both have raised before. Both have shipped real software at scale. Investors don't need decks for that.

Spend ten minutes with any engineering team in 2026 and you'll hear the same complaint. AI is generating code faster than ever. Reviewing it isn't getting any faster. Tests aren't getting any faster. Deployments are still queued behind the same brittle pipelines they were three years ago.

That's the gap Avrea is aiming for. The product, per the company's stealth-exit materials, plugs into existing deployment platforms and uses AI agents to spot the things slowing teams down. Low-quality tests. Stuck builds. Resource bottlenecks. The kind of stuff that gets noticed only when somebody loses a sprint to it.

There's a slightly uncomfortable irony here. AI coding tools like Copilot, Cursor and Claude Code are partly responsible for the bottleneck Avrea wants to fix. Generate three times as much code, ship three times as many bugs, queue three times the test runs. The pipeline can't take it. Teams either dilute review standards or watch deployment frequency collapse.

Valtonen's pitch is that the answer isn't to slow the AI down. It's to rebuild the pipeline around the assumption that most code is now AI-generated. Different failure modes. Different quality signals. Different tooling needs.

GitHub and GitLab, Avrea's largest competitors, were both built in the late 2000s. Their core product surface was designed for a world where software was written by humans, reviewed by humans, and shipped a few times a day. They have evolved. They have not, in any meaningful way, been redesigned.

That's a quiet but real opportunity. CI/CD as a category has been remarkably stable. Nobody has been able to dent the duopoly because the switching costs are awful and the incumbents keep adding features at a pace that just barely tracks the market. AI coding might be the first wedge in a decade where teams have a genuine reason to look at something new.

It's also the kind of opportunity that makes seasoned investors twitchy. Plenty of startups have promised to disrupt CI/CD. Most got swallowed, sidelined, or pivoted into adjacent categories. CircleCI, Buildkite, Harness, Octopus Deploy, plus a long tail of YC-funded experiments. The graveyard is well-tended.

Avrea's argument is that this time the platform layer itself is changing, not just the workflow on top of it. If true, that's the rare crack that lets a new entrant gain meaningful surface area before the incumbents can copy.

The lead, Earlybird, isn't an obvious fit on the surface. They're best known for European deep tech bets and the occasional growth-stage rocket. But Earlybird has been quietly rebuilding its developer tools thesis for two years now. This round is consistent with that.

The lack of public detail on follow-on investors suggests this was a tight syndicate. That's normal for repeat founders writing their own terms. It's also occasionally a flag that the round closed faster than the syndicate did, leaving angel slots that didn't get filled. Either way, expect a longer cap table by the next round.

There's a particular flavor of founder folklore that goes: we didn't need a deck, the round wrote itself. Nine times out of ten the truth is messier. There was a deck. It got shown to a handful of investors. Then a credible lead committed and the deck stopped mattering.

In Valtonen's case the no-deck claim is at least defensible. Aiven was a meaningful exit. The Helsinki investor circuit knows him. He could probably raise pre-seed money on a phone call. The interesting question isn't whether Avrea raised cleanly. It's what the round size signals.

$4.7M is unusually large for pre-seed. It's a Series A in some markets. It's also exactly the size of round that gives a repeat founder enough runway to ship a real product without the dilution of a proper Series A. Read it as a vote that Avrea's team can build a defensible v1 before they need to take a priced round. That's a smart structure if you can pull it off.

Some hard questions for the team. First: most enterprise CI/CD pain isn't actually about test speed. It's about test reliability, environment drift, and bad data. AI agents that flag bottlenecks are useful, but they're a long way from solving any of those harder problems.

Second: incumbents move slowly, but they do move. Microsoft has the budget, the data, and the relationship with every developer on the planet to bolt AI-driven pipeline analytics onto GitHub Actions in a release cycle. The window where a startup can outrun that is shrinking.

Third: the customer who needs Avrea most is the same customer who's hardest to sell to. Mid-to-large engineering orgs with mature CI/CD already. Those teams have invested years tuning their existing stack. Switching costs are high. Buying decisions are political. And the people doing the buying are exactly the kind who like to test their tools by breaking them.

None of these are deal-killers. They're just the hill the team has to climb. Valtonen has climbed similar hills before. He'd know.

Three signals will matter. The first is which teams sign up for design partner status in the next quarter. If it's a handful of well-known scale-ups with credible engineering reputations, that's a strong endorsement. If it's a parade of unfamiliar logos, the case is weaker.

The second is how Avrea positions itself relative to GitHub Actions and GitLab. Does it sit on top of those platforms? Replace them? Both? The answer changes the size of the addressable market and the difficulty of distribution.

The third is talent. Avrea's existing team draws on Spotify and Hoxhunt. That's a credible bench for Helsinki, but a 10-person team can't ship a CI/CD platform alone. The pace and seniority of the next ten hires will tell us how serious the company is about scale.

None of this is to say Avrea will or won't work. Pre-seed bets like this are properly speculative. But the founder pedigree, the round size, the timing of the AI coding wave, and the structural weakness of the incumbents all line up in the same direction. That's about as much as you can ask for at this stage.

Helsinki has a long, slightly underrated history of producing infrastructure companies with global ambition. Aiven came out of that culture. So did MariaDB, F-Secure and a handful of others. If Avrea joins that list it'll be partly because it caught the AI coding wave. And partly because the Finnish capital quietly, finally, learned to believe in pre-seed at growth-round prices.

Watch any startup ecosystem long enough and you'll spot the same pattern. The first generation of founders builds the companies. The second generation builds on top of them. Helsinki is now firmly in its second generation, and the data shows up in odd places.

Most of the strongest seed and pre-seed rounds out of Finland in the last twelve months have gone to founders who came up inside Aiven, Supermetrics, Nosto, Wolt, or one of the other 2010s-era successes. The networks compound. The capital follows the networks. The advice from a few cohorts up keeps getting cheaper.

Avrea is the latest data point in that pattern. Whether the company succeeds or not, the round itself is a useful indicator that the Finnish founder pipeline is producing repeat-founders at a steady clip. That matters for the ecosystem regardless of how Avrea's specific bet plays out.