TranscriptIntel exists because qualitative research teams deserve better tools. Not another dashboard with charts. Real analysis infrastructure.
TranscriptIntel is built by Gedank Rayze, a software engineering company based in Lisbon, Portugal. We build tools for teams that work with conversation data at scale.
Our focus is on the infrastructure layer: the runtime, the analysis pipeline, the data model. We believe the best tools disappear into the workflow and let researchers focus on insights, not software.
The experience that shaped the platform.
The founding team includes a former Cellebrite engineer with deep expertise in digital forensics and conversation intelligence. That background shapes everything about how TranscriptIntel handles data: chain of custody matters, evidence traceability is not optional, and analysis must be reproducible.
Working in forensics taught us that conversation analysis tools need to work across domains. The same patterns that matter in an interrogation transcript matter in a patient interview or a consumer focus group. The behavioral signals are universal; only the terminology changes.
Where we are and where we are going.
No hardcoded industry logic. Behavioral segments, analysis flows, and generation configs adapt to any vertical. Forensics, pharma, market research, HR, legal, education.
8 agents are live in production, 1 partially implemented, and 29 on the roadmap. Each agent is purpose-built for a specific analysis task.
We show what is implemented, what is partial, and what is planned. No vaporware. The agent roadmap on our Features page has real status badges, not marketing promises.
Purpose-built tools for conversation intelligence.
Multi-agent orchestration runtime written in Rust. Stateless compute with SSE streaming, per-run abort, model routing, and prompt caching. Coordinates multiple LLM calls into structured analysis flows with error recovery.
DAG-based workflow automation engine in Rust with Lua scripting. 96 built-in nodes for HTTP, file I/O, database, AI, and document extraction. Runs complex multi-step pipelines with parallel execution, retry logic, and conditional routing. Ships as a single binary.
Rust CLI for local-first transcription with speaker diarization and VAD segmentation. Supports whisper.cpp, sherpa-onnx, OpenAI, and Azure providers. Audio never leaves your machine. Any audio/video format via FFmpeg.
Cognition-aware document extraction engine in Rust. Transforms PDF, DOCX, PPTX, and images into structured Markdown using local ONNX models for layout detection and vision-language models for semantic understanding. Triages documents first, then applies the right extraction strategy per page.
Whether you are evaluating tools for your research team or want to discuss a partnership, we would like to hear from you.