Every question, answered

dembrane makes it easier to make sense of big, messy conversations. You bring the room together; we record every conversation at once, show live results while everyone is still in the room, and deliver an outcome document where every insight traces back to what was actually said. It comes as a self service platform, or as a hosted event where a dembrane host runs the session with you.

Participants scan a QR code with their own phone and start talking, in small groups or one on one. dembrane transcribes every conversation live, in any language. While the room is still together you see what is coming up across all the tables, and groups can check and refine their own summary. Afterwards you get a report in which every insight links to the quote, and the audio, it came from.

We specialise in empowering huge groups all at once.

Try getting 50 conversations on the record with a standard meeting notetaker - we have tried, it's a huge hassle.

Because we are specialised, we focus on all kinds of useful details: Built in pseudoanonymisation, Scan to record at scale, monitoring, dynamic dashboards. You name it. Even little details like: ChatGPT answers from the whole internet; dembrane answers only from what your people actually said.

No. Any phone with a browser is a recorder: participants scan a QR code and talk. Nothing to install, nothing to hand out.

More than 100 languages and their dialects: Dutch, English, German, French, Luxembourgish, Basque, Catalan, Welsh, Arabic, Turkish, Polish, Ukrainian, and a room can switch languages mid-sentence (a citizen assembly in Luxembourg runs in four at once). Because we refuse to cut corners on transcription, clients tell us it holds up where other tools slip: jargon, local terms, regional accents and mixed-language conversations.

As many as you like: every plan comes with unlimited participants. Around 500 people at once is where an event becomes a bigger programme, which gets a custom quote in the intake call.

Because you can check, and so can the room. With Verify, each group gets a first version of their own insights back on the spot and refines it by talking on, or typing, until it feels right; that verified record then weighs heavier in the analysis. At events, live feedback on screen shows whether people recognise themselves. And insights carry quotes straight from the transcript, with the audio behind them unless you run in full anonymous mode.

Of course! You can use dembrane to record and understand any conversation, but it shines at scale! Why have endless meetings when you can host the most impactful moment of the year and completely transform your organisation?

Yes, and you should. Every project has a context field for your goals, the spelling of names, and the themes you care about; the analysis uses it to frame what it reads. Every view starts from your own research question, and you can hand it a framework to cluster by. There are no default questions, because every session is different. The question participants see on their screen is context too: the framing of the session, visible when they leave the main room.

You can add text to a project the same way you add a recording: paste a report section, an email or a summary, and it becomes part of the data. A PDF upload is deliberately not there yet. We advise keeping a project pure and running the analysis on the conversations first, because documents added up front colour what comes out. The pattern that works: ask dembrane what the framing of a document is, and put that in the project context.

It is designed for rooms, and that is where it is unique. There is no Teams, Google Meet or Zoom integration. For an online session, a phone next to the speaker records it fine, and audio recorded with another tool can be uploaded to the dashboard afterwards. For video deliberation we are integrating Frankly, an EU-hosted platform built by a Harvard and MIT lab, with dembrane.

The analysis runs on the text, with context: not word by word but by what a passage points towards, so optimism and frustration are usually readable. There are no visuals, so body language is not captured, and we are experimenting with analysis directly on the audio. Because an AI can misjudge a room, the result goes back to participants to check and correct before it counts.

Yes. Most clients run each session as its own project, for a report per session, then merge projects for one big analysis. Within a project you can chat with specific recordings ("what are the biggest differences between Breda and Tilburg?"), and tags let you mark tables, locations or themes and filter the analysis by them. Asking one question across all your projects at once is being built.

Nothing is lost. Recording uses very little data, and a phone keeps recording for about fifteen minutes without a connection, then uploads when it is back. You can also record locally with any device and upload the audio to the dashboard afterwards.

Participants mostly just talk. At the end of a conversation they see a first summary and can say what is wrong or missing, which is how the output gets verified. If you want more, there is a button you can switch on with your own prompt behind it: a brainstorm partner, a blind-spot finder, or a Socratic question that takes the table deeper. Afterwards, participants who leave an email can receive a short summary of their own contribution.

Three things. The full transcripts, live while the session runs. A report, written as neutrally as possible from everything that was said and the context you gave, structured around the questions the conversations actually answer. And a chat with your own data: ask where the biggest tensions were, how much depth each theme got, which tables discussed what, and get tables and heatmaps alongside the text.

Mostly, yes. Be upfront about why you record: when stakes are high, people are often reassured to see every word carried exactly rather than a note-taker’s paraphrase. And participants hold the controls: there is a pause button on every conversation, because the point is not that everything is recorded, but that everything people want to share is. Our longest-running deployment, a people’s panel in Northern Ireland, saw more openness, not less.

In the Benelux, yes: a dembrane host (or even one of our founders, let us know) runs the session with you, from intake to report. In the UK, US and Australia, we work with partners. Everywhere else, your own team runs the room while dembrane listens live in every language and we provide remote event support. Not sure what you need? Get in touch.

We provide training as part of our EU AI Act risk mitigation strategy. For some use cases in the EU, including when dembrane is used in healthcare, infrastructure management, justice, education and/or HR contexts, training is mandatory. We strongly recommend training your team so that they know how to deal with otherwise difficult situations as they might arise when working with participants.

Follow a training and find out: two hours online for a single facilitator, or an afternoon for a team, and you leave knowing how to design a session, what to ask the data, and how to handle the difficult moments at a table. In the meantime, read how dembrane adds value.

Rolling out takes a day: an account, one sitting together to design the sessions, and we are set. We have run events planned five minutes in advance. What makes the difference is one session with the core stakeholder group that owns the outcomes, together with the day's chair. Keep individual recordings to about 45 minutes with a pause in between, and the QR code stays open for as long as you want input.

A hosted event is €1,250 for preparation (intake, event design, workspace setup) and €1,250 per event day, with consultation, event design, dedicated support and server queue priority included. Bigger programmes get a custom quote in the intake call.

Self service plans start at €75 a month with unlimited participants and unlimited recording hours.

Please do. A free account comes with your first hour of transcription, and a single seat already carries a full event: everyone scans the QR code, one person holds the dashboard. Most clients ran one session first: a team day, one table, one round of street interviews. Or bring us along to your next event and get expert care.

We never want money to be the reason you can’t listen to your community, but we also never want money to be the reason dembrane goes under. If you need budget assistance, you can apply for a scholarship, up to 100%.

Encrypted on EU servers. dembrane is ISO 27001 certified and GDPR-compliant, we have explicit agreements with every provider to keep processing on European infrastructure, and we never train models on your data unless you opt in. Your data stays only as long as you keep it: delete it and it is gone from our servers too. Everything is documented in the trust center.

Yes, you can find our source code on GitHub at github.com/dembrane/echo. Our platform is licensed under the Business Source License (BSL 1.1), which means you can deploy and use it yourself, unless you are using it on behalf of an organisation that makes more than 1 million euros a year, in which case we ask you to buy a commercial licence.

No, on purpose. Telling voices apart with any certainty is hard, and more importantly it is not what matters in a room: what was said counts, not who said it. The transcript has no speaker labels, personal details are removed from it, and the audio is deleted soon after. Sometimes the structure of a conversation (a question, then an answer) still makes it readable.

Yes. Accounts are personal, one email each, and a workspace can be shared with colleagues, who can then edit everything in it, like a shared document: no bottleneck through one owner. For sensitive projects the higher tier adds audit logs, which is why everyone should be in with their own address. Agencies make a workspace per end client, with its own billing.

Transcripts can be downloaded from the dashboard at any time, also when a subscription ends. Anything the chat produces can be copied straight into your own document. A direct export of views and reports is being worked on.

Mostly the public sector: municipalities such as Den Bosch, Zaltbommel and Amsterdam, ministries, VNG, provinces, and the NGOs and facilitators who run participation for them. Beyond the Netherlands we support a citizen assembly in Luxembourg that runs in four languages at once, and peacebuilding dialogues in Northern Ireland. dembrane started in 2022 in the public domain, and what every client has in common is that dialogue is central to how they work, and something has to be done with what people say.

Any moment where many conversations happen at once and you need to bring them together: citizen assemblies, participation evenings, multi-stakeholder sessions, workshops with 50 or 100 people, street interviews, and audio surveys where people talk instead of type. It shines when you want to see where conversations agree, where the biggest tensions sit, and what gets lost in the usual report of an atmosphere and a few photos. For a single meeting, see the question about meeting notes.

Yes. There is a case study on the peacebuilding work in Northern Ireland, written by People Powered, cases from Den Bosch and a university's governance process, a menu of what we offer, and a short video of what participants see. Book an intake call or email [email protected] and we send what fits your audience.