From raw data to reliable annotations.

AI-assisted annotation that you stay in charge of

Joey, Tagaroo's annotation agent, reads your data and drafts annotations against the scheme you chose: labeled spans, a severity rating, and a written rationale anchored to the exact evidence. You review each proposal — accept, edit, or reject — so the finished dataset is human-verified, not machine-guessed.

The agent flags what it is unsure about instead of forcing a label, and it sharpens as you correct it. AI-assisted qualitative coding here means the grunt work moves to the machine while the judgment stays with you.

A transcript annotation tool built for interviews

Paste plain text, import structured JSON, or transcribe audio and video in place — interviews arrive as speaker-turn transcripts ready to code. Select any span to tag a phenomenon, attach a severity from the scale's anchors, and keep the rationale with the annotation.

Interview coding that used to mean colored highlighters and spreadsheet cross-references becomes a single auditable workspace: every rating traces back to the utterance that justified it.

Validated clinical rating scales, ready to apply

The Scale Library ships 25 curated instruments — PHQ-9, GAD-7, HAM-D, MADRS, YMRS, BPRS, MITI, the NICHD protocol, Ekman emotions and more — each with its items, severity anchors, and source citation. Apply one to a study and it is immediately a working annotation scheme, not a PDF you re-type.

Score from the transcript rather than from memory: evidence-anchored ratings are what make clinical interview annotation auditable and defensible in a paper or a chart review.

Turn any codebook or scale into a live workflow

Upload a scale, codebook, or annotation guideline — PDF or text — and Tagaroo extracts its structure: phenomena, definitions, severity levels, decision rules. Review the extraction, adjust it, and start annotating with it in minutes.

Nothing about your coding scheme has to fit a template; if a human annotator could follow the document, the workspace can host it.

Image annotation in the same workspace

Draw regions on medical scans, documents, or photographs and tag them with the same schemes, severities, and rationales you use for text. One workspace and one export format across modalities means one QA process for a mixed-media study.

Multi-annotator studies with inter-rater reliability built in

Run studies with several annotators — human, agent, or both — and get agreement statistics without exporting to a stats package. A compare-and-consolidate view shows every disagreement span by span, so adjudication is a workflow, not an afterthought.

For the statistics themselves, the free Materials tools compute Cohen's kappa, weighted kappa, Fleiss' kappa, Krippendorff's alpha, Gwet's AC1 and ICC with confidence intervals — in the browser, no account needed.

Built-in transcription for recorded sessions

Upload audio or video and Tagaroo produces a speaker-attributed transcript you can annotate immediately. Recorded interviews, therapy sessions, and focus groups go from file to coded data without a third transcription tool in the loop.

Privacy-first, EU-hosted, consent-only analytics

The no-signup trial runs entirely in your browser — trial data never leaves your device. Workspace data is hosted in the EU and processed under GDPR; identifiable clinical data is not permitted, and a free in-browser transcript de-identifier helps you strip names, dates, and places before anything is uploaded.