A five-step pipeline that takes any conversation recording and produces structured, quantified analysis. Every step is observable, every run is cancellable.
Each step runs in sequence. SSE streaming shows progress in real time.
Upload VTT, SRT, or plain text transcripts up to 50MB. Drag-and-drop or API upload. File extension and content signature are both validated.
Automatic format detection identifies the file type by signature, extension, and content analysis. Speakers are extracted, timestamps normalized, and the transcript is stored in a canonical schema.
Select from 7 analysis types. Each runs as an IronCrew flow with one or more AI agents. Results stream back via SSE with per-phase progress updates.
Examine structured results: summaries, topic clusters, extracted quotes, sentiment timelines, speaker profiles, behavioral DNA classification, and interviewer scorecards. Every finding links back to source evidence.
Export analysis results as JSON for programmatic consumption, CSV for spreadsheets, or formatted Markdown for reports and presentations.
Built-in format detection handles the messy reality of transcript files.
WebVTT format with full timestamp and speaker label parsing. Handles multi-line cues, notes, and style blocks.
SubRip subtitle format. Timestamps are converted to the canonical schema. Speaker labels extracted from cue text.
Unstructured text with speaker prefixes (e.g., "Speaker 1:"). Heuristic detection identifies speaker turns without timestamps.
Need to transcribe audio or video first? The transcribeit CLI handles local-first transcription with speaker diarization. Rust-based, runs entirely on your machine.
Need document context? Extract structured content from PDF, DOCX, PPTX, and images. Use as additional context for generating new or analyzing existing transcripts.
Watch analysis unfold in real time. Cancel any run mid-flight.
Every analysis streams progress via Server-Sent Events. The UI shows each agent phase as it completes, with inline results and timing information.
No polling, no page refreshes. The connection stays open from start to finish, delivering structured events for each phase transition.
Get results in the format your downstream systems need.
Full structured output with all analysis fields, confidence scores, and source references. Ready for programmatic consumption.
Flattened tabular output for spreadsheets and BI tools. One row per finding with all metadata columns.
Formatted reports with headings, tables, and inline quotes. Ready for presentations, wikis, or documentation.
Upload a transcript and watch the full analysis pipeline run. We will walk you through every step.