Nordic Makers has invested in Good Tape, the Copenhagen-based AI transcription company used by 3 million people in more than 150 countries. Alexander Aghassipour, co-founder and former Chief Product Officer of Zendesk, led the investment for Nordic Makers and joined Good Tape's board.
Good Tape started as an internal tool at Zetland, the Danish member-funded newsroom.
Journalists built it because they needed accurate transcripts without handing sensitive interviews to third parties.
Led by co-founders Ýmir Gyðuson Gíslason and Lasse Finderup, it has since completed more than 15 million transcriptions in over 100 languages for journalists, researchers, consultants, lawyers, and public-sector teams.
Today Good Tape launched Explore, which connects every new recording to everything a user has transcribed before.
It surfaces changed opinions, contradictions and supporting arguments, each linked to the original recording.
Journalists, researchers and teams can trace what was said, when, and follow developments across months of recordings.
I spoke to Lasse Finderup, co-founder and CEO, Good Tape to find out more.
"Good Tape was built inside a newsroom by people who needed to trust where their interviews ended up. Zetland gave us the room to build it, and we owe them a lot.
Now we're going all in on that trust. With Explore, Good Tape goes from transcribing your interviews to also connecting them, and it's still safe for your most sensitive material," said Finderup.
Why Good Tape bought out Zetland
The investment was part of a management buyout that made Good Tape an independent company. Finderup took majority ownership, buying out the stake held by Zetland, the Danish media company from which Good Tape was spun out. With the buy-out, Zetland is no longer an owner.
Finderup said the previous structure had become an obstacle when the company sought external investment.
“When we wanted to raise investment, it became clear that we couldn't really do that as a spin-out because no one wanted to invest in that. We solved it with a management buyout.
Raising a round is a big deal in itself, but doing a management buyout and raising a round on the same day took some structuring.”
The deal actually ended up going through on his wedding day. Since then, Good Tape has largely been in what Finderup describes as “building mode” — talking to customers and developing the product — culminating in this week’s launch of Explore.
Check out an earlier interview with Finderup.
From transcription to memory
Like a lot of transcription services, Good Tape realised around one and a half to two years ago that transcription was going to become a commodity.
“You have to do something more than just transcription.”
Explore is a context layer that sits on top of everything you've transcribed.
“What we see today is that a journalist might do an interview like this and then use five to 10 per cent of it — a great quote, something for the headline — and then never go back to it. You don't really have time to go back, but there is so much gold just sitting there. What we've built is a way to be reminded of things you've forgotten,” said Finderup.
It's a useful proposition for a journalist like me who often interviews a company early in its journey — perhaps around its first raise — and then returns as its product offering matures and its team expands. Finderup explained:
“For example, if I say something today that contradicts something I told you two years ago, Explore could surface it and say, ‘He said something different.’”
Crucially, Explore lets users check whether someone has changed their opinion from an earlier interview or return to the original recording if a source disputes what they previously said.
“You can already hack your way towards some of what this will be able to do. You might find that someone said something different half a year ago, or that you have an expert who supports what they're saying. You can play it back and listen to it immediately.”
Making AI fit the journalist’s workflow
Good Tape also offers speaker labelling and is working on voice fingerprinting, which Finderup says will eventually remove the need to manually identify speakers. The idea is to reduce the administrative work involved in organising interviews and remembering what belongs to which project. The company has also simplified its UX with a Recorder app built around what Finderup describes as “a big red button.”
“What I always say to buyers in companies is that you don't have to teach people all these crazy things. They click the big red button, and then they click the big red button again. It's designed to be very simple.”
For journalists hesitant about adopting new AI tools, Finderup said the combination of the Recorder app and Good Tape’s MCP has helped make the workflow easier to grasp.
“I show them: click the big red button, click the big red button, and then go straight to Claude, Copilot, GPT, or whatever, and do your thing. That's where I see people go, ‘Okay.’”
It also changes Good Tape’s original workflow. Rather than recording, uploading, editing, and then exporting a transcript, users can move from recording directly into the LLM they already use.
“If you find something while you're working there, you go back and edit. You might need to check a quote, listen to the audio, or edit a name.”
He describes Good Tape’s original proposition as saving journalists time.
“We did that, and now they're much more efficient than they were.
The next step is allowing them to create deeper quality in the same amount of time, because they don't have more time.
You can write incredible articles if a company says, ‘I'll pay you for half a year to go in depth on this.’ You can go super deep. But that's not the reality if your deadline is tomorrow and you also need to read another story for the day after.”
Who controls the organisational memory?
Explore can also extend beyond an individual journalist's archive.
“You can be reminded of something I told your colleague, if they have shared it. Then you can start to create what we would call an organisational memory.”
But that introduces another issue: who gets access to an interview once knowledge is shared across an organisation?
Finderup admits, “I also know journalists well enough to know that everyone feels, ‘No, that's my interview.’”
He saw that tension first-hand during a recent visit to Helsinki, where he met with 12 of the country’s 20 largest media companies. In meetings involving both journalists and heads of AI or IT, he said the difference in response was clear.
“The head of AI immediately went, ‘Yes,’ while the journalists were much more hesitant.
We need to keep in mind that the user will always be the journalist. We need to give them control so they feel safe and will use the product, and then the other benefits will become clear through use.”
Some journalists have also raised ethical questions around resurfacing old material. For example, a source could potentially be quoted in a new article using something they said in an interview two years earlier that was never published. Finderup argues that existing journalistic judgement still applies:
“What I say is that it's the same as it is now. If you feel like you should get that quote approved, then you still do it.”
Good Tape’s customers include organisations such as Human Rights Watch, where privacy is critical and multiple people may work on the same subject over long periods — a use case where Finderup says Explore is particularly valuable.
The product carries Good Tape’s existing protections, including ISO 27001 certification, GDPR compliance, EU data storage, and a commitment not to train AI models on customer data.
That focus on trust and journalist safety extends beyond the product itself. In June, Good Tape launched its Names campaign, highlighting journalists killed in the line of duty. The initiative also included the Good Tape Grant, offering up to €500 to independent journalists and storytellers, alongside support for organisations working on journalist safety, mental health, and professional development.
Turning an interview archive into context
Explore can organise previous conversations into projects. For example, a user could ask it to create a project containing everything they have discussed with a particular person. The platform can then identify the main topics and search the wider interview archive for relevant material.
Finderup explained:
“Imagine you and I are chatting and this is our conversation. Explore can look through the archive to see whether there's anything relevant to the conversation we've just had.”
One of those connections could be another source offering a different perspective: “It might be that an expert has said the opposite of what I just said.”
For Finderup, the aim is to help journalists draw connections between interviews that might otherwise remain buried in their archives.
“My hope is that every article can become less ‘he said this’ and more ‘he said this, but…’ or ‘he said this, and…’. You can add that extra layer that makes the article deeper.”
From artificial memory to real-time context
Good Tape is also seeing demand from consultants and researchers — people who, like journalists, have large numbers of conversations and need to recall information across them. Finderup sees the same problem in his own work:
“It's also useful for people whose memory is under pressure, my own included. I context-switch 10 times a day. To be able to walk into a meeting and know exactly the number that you told me in November can be incredibly helpful.”
The more ambitious next step is bringing that artificial memory into conversations as they happen.
I asked Finderup whether Explore could eventually work in real time — for example, surfacing during an interview that the person on stage had said something different the previous week.
“If you're interviewing politicians, for example, you could say, ‘You said something different before — let me play it back to you.’”
For Finderup, that could change the dynamic of interviews:
“It could create a world where people have to stand by what they've said.”
For now, however, Good Tape's priority is getting people to use Explore. Most of its users currently arrive organically, particularly through word of mouth.
“What we often see is that if we get one user in a company, there will probably be another, and then probably another. At some point, I'll be talking to the head of IT.”
The company also isn't initially charging users for Explore.
Write Comment“We're not even going to charge anything initially. It's about getting people to use it and have those ‘wow’ experiences. “At some point, we'll say, ‘Now it's going to cost something.’”