# dembrane

> Turn complex conversations into common ground with the stakeholder engagement platform built for the real world.

dembrane is the stakeholder engagement platform that proves you are really listening — and helps you take action with trust. It turns dozens of small-group conversations into shared outcomes using live transcription, language model analysis, and traceable reports, all while the room is still buzzing.

- Vision: A world where the more people talk together, wisdom, impact, and connectedness grow.
- Mission: Empower every voice to participate in decisions that impact them and their community.
- Tagline: Together we make sense.

## Company Facts

- Name: dembrane
- What does dembrane mean? "democratic membrane"
- Founded: 2023 (rooted in the founders' master's research — Jorim Theuns (2020) and Evelien Nieuwenburg (2023))
- Founder: Jorim Theuns — Head of Systems and Design. Conceived dembrane in 2020 from his master's research on social media's effects on democracy and deliberative design.
- Co-founders:
  - Evelien Nieuwenburg — Head of Value Delivery (joined 2022)
  - Bram Delisse — Head of Administration (joined 2023)
  - Sameer Pashikanti — Head of Engineering (joined 2024)
- Founded in: Eindhoven, Netherlands
- Based in: 's-Hertogenbosch (Den Bosch), Netherlands
- Governance: No CEO — the founding board votes collectively on crucial decisions
- Team: ~8 FTE, based at the Marienburg campus
- Industry: Participatory technology / Civic tech / GovTech / DelibTech
- Target users: Governments, municipalities, consultancies, NGOs, conference organisers, community networks
- Website: https://dembrane.com

## Traction

- 3,000+ hours of conversation recorded
- 13,000+ conversations analysed
- 1,500+ projects
- Trusted by the United Nations, OECD, VNG, and the Dutch Ministry of Social Affairs
- 100+ organisations across Europe

## What dembrane does

dembrane is one sealed flow from the first spoken words to verified, publishable insights:

1. **Talk.** Participants scan a QR code and talk. They can share individually or have natural face-to-face discussions while phones quietly capture audio at the table.
2. **Listen.** Secure language models transcribe and synthesise core themes, hidden tensions, and breakthrough ideas from dozens of tables simultaneously. EU-hosted. 150+ languages.
3. **Verify.** Before the session ends, participants check the output on their screens. They confirm, refine, or reject: "Yes, this is what we meant."
4. **Understand.** Leaders receive a structured, participant-backed report — fully traceable to source quotes without breaching trust.

Setup takes less than five minutes. No app download needed.

## Key Features

- QR-code-based participation — no app downloads, works on any smartphone
- Real-time transcription in 150+ languages with excellent first-party support for all European languages
- Interface available in 6 languages: English, French, German, Spanish, Italian, Dutch
- The ECHO button: empower participants with tools for exploring new perspectives and verifying outcomes.
- Live host dashboard showing what's emerging across all conversations
- One-click structured reports with traceability back to source conversations
- Export to PDF and Markdown & web hook support.
- Argument mapping (coming soon)
- Whitelabeling available to partners

## How dembrane is different

- dembrane captures dialogue, not monologue. Unlike surveys, Mentimeter, or Polis, it's built for actual human conversation
- Face-to-face first: technology disappears into conversation rather than replacing it
- Trust-centred: participants verify output, we encourage a critical stance toward AI.
- Does not replace facilitators — gives them superpowers. Partners with expert consultancies and process guides
- One sealed flow: every handoff in the DIY approach leaks data, loses participants, and breaks compliance. dembrane keeps the chain intact from voice to verified insight

| Alternative | How dembrane differs |
|---|---|
| Surveys | Captures conversation, not isolated responses |
| Mentimeter/Slido | Goes deep on dialogue, not surface polling |
| Polis | Face-to-face first, not async-only |
| Transcription tools (Copilot, Otter, etc.) | Built to empower many participants with interactive conversation support, live. Makes sense of context, not just text. Data stays in EU, participants verify output, data can be anonymised at source |
| Generic AI (ChatGPT, Copilot) | Built to foster trust, not generate text. DIY pipelines break compliance and leave participants out of the loop |

## Why now

- Language models make real-time, multilingual, anonymised conversation analysis possible for the first time
- EU regulation increasingly requires structured participation — and funds it (€9B Democracy Shield / AgoraEU)
- Growing demand for face-to-face connection and solving real problems together — dembrane meets that need

## How the technology works

dembrane uses a two-model transcription stack that delivers state of the art results, especially for mid sentence language switching. Our approach is shown to reduce hallucinations and improves handling of accents and dialects. All processing happens on Northwest European servers. dembrane is not locked into a single provider.

The ECHO button has two modes:
- Explore: the language model reads your transcript and conversations at nearby tables, then offers a Socratic question to deepen the discussion
- Verify: participants choose what to surface (hidden gems, agreements, disagreements). The model generates output, participants critique it, then verify or refine

## Trust & Compliance

- ISO 27001 certified
- GDPR compliant
- EU hosted (Northwest European servers)
- Source available on GitHub (BSL 1.1 license — visible and auditable, not permissive open-source)
- Anonymous by default (Chatham House rules)
- No training on client data
- Data deleted 30 days after project ends (or immediately on request)
- Sub-processors: Assembly AI, Google Cloud Vertex AI, Azure, Runpod
- Using AI in participatory settings is classified as high-risk under the EU AI Act — dembrane requires mandatory training for all teams

## Security Tiers

- Baseline: EU servers, European-hosted AI models (default)
- Sovereign: Full EU tech stack (OVH hosting, Mistral AI models)
- Maximum Control: Self-hosted, on-premise, local LLMs

## Pricing

### Event Services (full-service, dembrane attends in person)
- EUR 1,250 event preparation (intake, design, workspace setup)
- EUR 1,250 per event day (in-person support, live analysis, structured report)
- Optional extras: argument mapping, quantitative analysis, additional facilitators
- Can handle events with up to 1,000 recording hours

### Platform Subscriptions (self-service workspace)
- Pioneer: from EUR 200/mo — 25 recorded hours, 3 seats, email support
- Innovator: from EUR 500/mo — 50 recorded hours, 10 seats, SSO, audit logs, phone support
- Changemaker: from EUR 1,500/mo — 100 recorded hours, 20 seats, whitelabel, API, dedicated event support
- Guardian: from EUR 5,000/mo — unlimited hours and seats, on-premise available, dedicated account owner
- All tiers: unlimited participants, EU hosting, GDPR compliance, ISO 27001

### Training (mandatory under EU AI Act for high-risk use)
- Online: EUR 675 (2 hours, up to 5 participants)
- In-person: EUR 2,500 (4 hours, up to 10 participants)
- Pilot: EUR 349 (1 month platform access + preliminary session for 2)

### Scholarships
- 25-100% discounts for NGOs, grassroots movements, students, and frontier use cases
- In exchange: community feedback, beta testing, or case study documentation

## Use Cases

- Citizens' assemblies (e.g., Derry: 100 citizens, 3 days; Flanders: 200 participants, same-day report for the Minister)
- Conferences and congresses (e.g., AMG: 130 people, 13 roundtables, synthesis delivered before departure)
- Stakeholder research and co-creation
- Community and neighbourhood dialogues
- Focus groups and foresight exercises
- Youth summits and peacebuilding debates

## Frequently Asked Questions

Q: What is dembrane?
A: A stakeholder engagement platform that turns dozens of small-group conversations into shared outcomes using real-time transcription, language model analysis, and traceable reports.

Q: How is dembrane different from Mentimeter, Slido, or surveys?
A: dembrane captures dialogue, not monologue. Surveys and polling tools collect isolated individual responses. dembrane listens to actual human conversation and makes sense of what was said collectively.

Q: Why not just use ChatGPT or Copilot to analyse transcripts?
A: Generic AI is built to generate text. dembrane is built to foster trust. The DIY approach — record on phones, paste into an LLM, hope for the best — breaks compliance and leaves participants out of the loop. With dembrane, data never leaves the EU, participants verify the output, data can be anonymised at source, and everything is traceable back to what was actually said.

Q: Does dembrane replace facilitators?
A: No. dembrane gives facilitators superpowers. It partners with expert consultancies and process guides, providing the platform they need to handle larger, more complex events.

Q: What is ECHO?
A: ECHO is dembrane's participant-facing language model feature. Explore mode offers Socratic questions to deepen conversation. Verify mode lets participants surface insights, critique them, and refine them collaboratively.

Q: How accurate is the transcription?
A: dembrane uses a two-model stack (AssemblyAI + Gemini) that cross-checks transcripts across two passes, reducing hallucinations and improving accuracy for accents and dialects.

Q: What languages does dembrane support?
A: 150+ languages for transcription with excellent first-party support for all European languages. The interface is available in English, French, German, Spanish, Italian, and Dutch.

Q: Is dembrane GDPR compliant?
A: Yes. ISO 27001 certified, GDPR compliant, EU hosted. Anonymous by default. Audio deleted 30 days after project ends or immediately on request. No training on client data.

Q: Who uses dembrane?
A: The United Nations, OECD, VNG, the Dutch Ministry of Social Affairs, and 100+ organisations across Europe. 3,000+ hours recorded across 1,500+ projects.

Q: How much does dembrane cost?
A: Event services from EUR 1,250 per event. Platform subscriptions from EUR 200/mo. Pilot at EUR 349. Scholarships available (25-100% off) for NGOs, grassroots, students, and frontier use cases.

Q: Can I export data from dembrane?
A: Yes. Reports and data can be exported as PDF and Markdown. API access is available on Changemaker tier and above.

Q: Where is dembrane based?
A: Founded in Eindhoven, currently based in 's-Hertogenbosch (Den Bosch), Netherlands.

Q: Who founded dembrane?
A: Jorim Theuns (Head of Systems and Design) founded dembrane based on his 2020 master's research. Co-founders Evelien Nieuwenburg (Head of Value Delivery), Bram Delisse (Head of Administration), and Sameer Pashikanti (Head of Engineering) joined between 2022 and 2024. There is no CEO; the founding board votes collectively on crucial decisions.

Q: Is dembrane open source?
A: Source available under BSL 1.1 — the codebase is visible and auditable on GitHub, but it is not a permissive open-source license.

Q: Does dembrane need an internet connection?
A: Yes. Audio is sent to EU-hosted servers for real-time transcription and analysis.

Q: Can dembrane handle multiple languages in one conversation?
A: Yes. A conversation hopping between five languages is not a problem.

Q: How long does setup take?
A: Less than five minutes. Participants scan a QR code — no app download needed.

Q: Can dembrane be self-hosted?
A: Yes, via the Guardian tier or by organisations under the revenue limit as stated by our BSL 1.1 licence.

Q: Does dembrane train on my data?
A: No. dembrane never uses client data for training models.

Q: What is Dembrane ECHO?
A: ECHO is the interactive button that empowers participants with language model capabilities like Verify and Explore. Formerly, Dembrane ECHO was the former name for the dembrane platform. Now we refer to the dembrane software platform broadly as dembrane. 

Q: What is the maximum event size dembrane can handle?
A: Up to 1,000 recording hours per event. Beyond 100 hours (e.g., 300 people deliberating for an hour in groups of 3), additional costs may apply.

## Contact

- General: https://forms.dembrane.com/contact
- Privacy: privacy@dembrane.com

## Key Pages

- Features: https://dembrane.com/platform/features
- Pricing: https://dembrane.com/platform/pricing
- Use Cases: https://dembrane.com/platform/use-cases
- Trust Center: https://dembrane.com/trust
- Case Studies: https://dembrane.com/case-studies
- Events Service: https://dembrane.com/events
- About: https://dembrane.com/about
- Blog: https://dembrane.com/blog

## Last Updated

February 2026