← All use cases
For researchers

Qualitative interviews, on-device by default.

Record participant interviews and focus groups locally with on-device Meeting Notes. Dictate methods, results, and discussion sections. Ask the AI chat about a paper or transcript on screen. No cloud routing of participant audio, IRB protocols intact.

Download for Mac

A field guide for researchers, updated May 4, 2026 by Branden Chong.

In practice

How researchers use Dollop.

  1. 01.

    Record participant interviews and focus groups.

    Hit ⌃ R when the interview starts. Meeting Notes captures audio (mic for in-person, system audio for Zoom/Teams), transcribes locally on Apple's Neural Engine, and produces a clean transcript with a thematic summary. Recordings and transcripts stay on your Mac, IRB protocols and participant consent terms intact.

    Heard

    themes from interview 7: participant consistently distinguished between trust in the platform and trust in individual creators. emergent code: 'platform vs creator trust'. cross-references to interview 3 and interview 5.

  2. 02.

    Dictate methods, results, and discussion.

    Hit ⌥ Space and talk through the methods section the way you'd describe it at a conference. Dollop's cleanup adapter holds technical vocabulary intact, and the per-app tone keeps Mail formal for advisors, casual for Slack, terse for Zotero notes.

  3. 03.

    Ask the AI chat about a paper on screen.

    Open the PDF, hit ⌃ S to summon the chat overlay with screen context, and ask for the methodology, the limitations, or a one-paragraph summary you can use in your literature review. Reads what's on screen locally; the paper doesn't leave your Mac, and the chat doesn't see anything you haven't shown it.

  4. 04.

    Coding sessions and team meetings.

    Record qualitative coding meetings, advisor sessions, or research team standups with Meeting Notes. Get the discussion as a transcript and the action items as bullets, all locally. Useful when the conversation references unpublished participant data.

What you actually get

Dollop mapped to your work.

Meeting Notes (⌃ R)
Records participant interviews and focus groups locally. IRB-friendly by architecture, no upload of participant data.
Voice dictation (⌥ Space)
Methods, results, and discussion sections drafted by voice in Word, Pages, Google Docs, Zotero notes.
AI chat with screen context (⌃ S)
Open a paper, ask for the methodology or limitations, locally. The PDF doesn't leave your Mac.
Custom domain vocabulary
Theoretical frameworks, methodologies, instrument names, and participant pseudonyms recognized verbatim.
Speaker turns preserved
Multi-person interview transcripts export with attribution, ready for downstream coding in Atlas.ti or Dedoose.
Works in fieldwork settings
Offline capable. Useful for remote sites, libraries with weak Wi-Fi, or institutional settings restricting cloud tools.
A type specimen

How the cleanup reads.

Real before-and-after dictation on terms specific to researchers, processed locally on Apple Foundation Models.

Field memo (theoretical vocabulary preserved)
Heard

uh the participant kept distinguishing between trust in the platform and trust in individual creators which maps onto the literature on epistemic injustice and bricolage. emergent code platform versus creator trust. cross reference interview three and interview five.

Cleaned

The participant kept distinguishing between trust in the platform and trust in individual creators, which maps onto the literature on epistemic injustice and bricolage. Emergent code: platform vs. creator trust. Cross-reference: interview 3 and interview 5.

Methods section (passive voice)
Heard

uh participants were recruited via purposive sampling from three online communities. semi structured interviews lasting 60 to 90 minutes were conducted via zoom and transcribed locally on the researcher's machine. transcripts were analyzed using reflexive thematic analysis.

Cleaned

Participants were recruited via purposive sampling from three online communities. Semi-structured interviews lasting 60-90 minutes were conducted via Zoom and transcribed locally on the researcher's machine. Transcripts were analyzed using reflexive thematic analysis.

Why it fits

Why qualitative researchers care.

IRB-friendly by architecture.

Most IRB protocols restrict where participant audio can be stored and processed. Cloud transcription tools (Otter, Trint, Rev) route audio through third-party servers, which can require a protocol amendment or outright disallow them. Dollop processes everything on-device on Apple's Neural Engine and Foundation Models, no upload, no third-party AI vendor.

Free, no per-minute or per-seat billing.

Rev charges $1.50/minute for human transcription, ~25¢/minute for AI. Trint and Otter charge per seat or hour. For researchers running 30-60 long-form interviews per study, that bill adds up fast. Dollop is free, including unlimited recording and transcription.

Custom vocabulary for your domain.

Add the technical terms, acronyms, participant pseudonyms, and study-specific vocabulary that off-the-shelf dictation gets wrong. Dollop AI learns them locally, no retraining anything cloud-side.

Works on a laptop, in a study room, in the field.

Recording, transcription, dictation, and AI chat all run on-device. No internet required, useful for fieldwork in remote sites, libraries with weak Wi-Fi, or institutional settings where cloud tools are restricted.

Asked & answered

Questions, answered.

Is Dollop IRB-compliant for participant interviews? +
IRB compliance is a function of your specific protocol, but the relevant architectural property is on-device processing: Dollop does not upload audio or transcripts to any third-party server. That removes the most common IRB concern with off-the-shelf transcription tools (cloud routing of participant data through unreviewed third parties). Some institutions still require an explicit protocol clause for any AI-assisted analysis; check with your IRB.
How does it compare to Otter, Trint, Rev, and Atlas.ti for qualitative work? +
Those tools upload audio to the cloud and most train models on user data unless you opt out. Dollop processes everything on-device. The trade-off: Dollop transcribes and summarizes; it does not yet do qualitative coding, theme extraction across studies, or shared codebooks like Atlas.ti or Dedoose. The natural workflow is Dollop for the recording and transcript, then your existing qualitative analysis tool for coding.
Does it handle long interviews (60-120 minutes)? +
Yes. Meeting Notes records as long as you want; transcription runs in real time, so the transcript is ready the moment you stop. The summary generates locally in seconds.
Can I add discipline-specific vocabulary? +
Yes. The Custom Vocabulary panel takes any term, theoretical frameworks (e.g., "epistemic injustice," "bricolage"), specialized methodologies, instrument names, even participant pseudonyms. The adapter learns them locally.
Does it preserve speaker turns for multi-person interviews? +
Yes. Meeting Notes outputs speaker-attributed transcripts so you can distinguish interviewer and participant in the export. Useful for downstream coding.
Can I dictate into Google Docs, Word, Pages, and Zotero? +
Yes, system-wide. Dollop pastes clean text at your cursor in any app, browser-based or native.
Does it work without internet? +
Yes, fully. Recording, transcription, and AI chat all run on Apple's Neural Engine and Foundation Models on-device. The only network calls are software updates.
Hardware requirements? +
Apple Silicon (M1+) on macOS 26 with Apple Intelligence enabled.

Your participants trusted you with their stories. The transcription tool you use should not introduce a third party your IRB never approved. Download Dollop and keep the recording, the transcript, and the summary where they belong: on your Mac.

Download for Mac
Dollop

The voice OS for your Mac. Built with care in California.

© Dollop 2026