Field capture, structured by AI
Diktafon is a mobile capture system that turns voice recordings, photos, and notes into structured, searchable data. A native app for hands-free capture. An AI backend for transcription, image analysis, and entity extraction. A web interface for review. Each part works independently — and the entire system can run on your infrastructure.
Diktafon is built as independent components with clear responsibilities. The mobile app captures. The backend processes. The web interface reviews. Each part can be updated, scaled, or replaced without affecting the others. AI is pluggable — cloud, on-prem, or off entirely.
iOS and Android. Record voice, take photos, add notes — one-handed, while you work. GPS and timestamps are captured automatically. Sessions queue locally and sync when connectivity is available.
Structures the raw capture, then enriches it with AI: speech-to-text, image analysis, entity extraction, and summarisation. Purpose-built for Swedish — speech recognition tuned for Swedish audio and accents, and entity extraction trained on Swedish context.
Review sessions, read transcripts, browse photos, play audio with waveforms, and see capture locations on a map. Trigger processing manually or let it run automatically. Built into the Digital Tvilling platform.
Pair devices via QR code — no sensitive credentials stored on the phone. If a device is lost or stolen, revoke its access remotely and it's locked out immediately.
Diktafon is designed for organisations where field data matters and data sovereignty is not optional.
Myndigheter, kommuner, and regioner with strict data sovereignty requirements. Swedish speech-to-text, Swedish entity extraction, and hosting entirely in Swedish data centres. The system speaks your language — literally.
Non-negotiable data perimeters. No commercial cloud AI. Air-gapped or VPN-only deployments. Diktafon runs fully on your infrastructure — from AI processing to storage — with no external dependencies.
Replace clipboards, spreadsheets, and email chains with structured digital capture. Walk-through inventories that used to take days become a set of sessions. Field data goes straight into systems — no manual re-entry.
Full control over where data is stored and processed. No AI calls leaving the perimeter. Revocable device access. Auditable endpoints. Integrates with your existing infrastructure and security model.
Diktafon runs wherever your requirements demand. The same system, the same capabilities, the same mobile app — on the infrastructure you choose.
Managed hosting in Swedish data centres with Swedish AI models — speech recognition optimised for Swedish and entity extraction that understands Swedish organisational context. We run the infrastructure, you use the system. The fastest path to production.
Deploy to your existing cloud environment — AWS, Azure, GCP, or others. Your account, your region, your rules.
Run the full system on-premises. Everything inside your network. AI processing stays internal — no data leaves your perimeter.
Fully air-gapped deployment. AI models run on local hardware with no external network dependency. For environments where nothing leaves the building — not even a network call.
Diktafon is designed for environments where data sensitivity is not optional. The system enforces separation by design: the mobile app never stores sensitive credentials. The processing backend runs behind your network boundary. AI is pluggable — use cloud models, on-prem models, or disable AI entirely.
The phone holds a revocable access token — nothing else. No keys, no passwords, no direct access to backend systems. If a device is lost, revoke its access remotely and it's locked out immediately.
AI processing can run entirely within your infrastructure — no external API calls, no data leaving your perimeter. Or use cloud models if that's what you prefer. The system adapts to your requirements.
Speech-to-text and entity extraction tuned for Swedish. Hosting options in Swedish data centres. Built with Swedish regulatory requirements and operational context in mind.
The mobile app queues locally and syncs when connectivity is available. The backend operates independently. No hard dependencies on proprietary cloud services or external APIs.
Clear separation between components. You know exactly what runs where. No black boxes, no hidden dependencies, no unclear data flows.
Integrates with your existing identity systems, network architecture, and security policies. Deploy behind VPN, in DMZ, or fully air-gapped.
Strict environment mode: For truly strict environments, Diktafon can run with AI processing entirely disabled — raw audio and images are stored, but no transcription, entity extraction, or summarisation happens. The system becomes a pure capture and retrieval tool, with no algorithmic processing of sensitive content.
This mode is designed for environments where even on-premises AI processing is prohibited — maximum control, minimum automation.
Visit the Diktafon site to learn more, request early access, or contact us to discuss your requirements.