QUICKWINS.AI/TOOLS/DOCS-KNOWLEDGE/DOVETAIL

Dovetail

BUILT-IN AI INVOKE-ONLY
← Docs and knowledge
25 of 51Reads only
YESBuilt-in AI
51Actions
0Event triggers

Dovetail is an Australian software company that provides tools for transcription analysis, coding interpretation of interviews, survey responses, and feedback, enabling users to create summarized insights from their research analysis.

What your AI can do here Connector snapshot 2026-08-23

Connect Dovetail to ChatGPT or Claude with your own Dovetail login or an API key, and your AI can run 51 of its actions. 25 of those only read and change nothing. 6 can delete or send something that cannot be undone, so those need a boundary you set before anyone runs them.

This describes what Dovetail supports, which is one half of the picture. Where you start depends on the task and on how much you have handed over before, and no page can tell you that from a count.

Things you could hand over

Checked 2026-08-22

Candidates, not recommendations. These are things Dovetail supports once connected. Whether any of them is right for you depends on your own work, which this page knows nothing about.

What have customers actually said about this? No prior practice needed

Dovetail exposes a search that runs across the whole workspace at once, so your assistant can ask one question and get back the highlights, notes, insights and tags that answer it rather than a list of projects to go and open. It can then pull the full text of anything it found as markdown, which means it can quote people accurately instead of paraphrasing from a title. Nothing is created, edited or moved.

What you end up with. The actual quotes on your question, with the note and project each one came from, in the words the customer used.

Start with it bringing you the information and you deciding.

Never without. Ask where each quote came from and check that the sample is not one project. Search returns what matches the wording of your question, so a theme that three people phrased differently can look smaller than it is.

Built on: Magic Search / List Highlights / Export Note / Get Note

Did we already research this before I go and run it again? No prior practice needed

Before scoping a study, the assistant can list the insights in the workspace, filter them by project or by whether they were published, and read the ones that look close to your question. Insights are the synthesized layer, so this is the fastest honest answer to whether the team already knows something. It reads and reports; it does not touch the record.

What you end up with. A short list of prior insights that bear on your question, each with its title, its project, and what it concluded, plus an explicit note of where nothing was found.

Start with it bringing you the information and you deciding.

Never without. Treat a nil result as unindexed rather than unstudied. Work that lived in a doc or a raw note and never got promoted to an insight will not show up in an insight listing at all.

Built on: List Insights / Get Insight / Export Insight / List Projects

Reading a stack of notes and telling me what the pattern is Assumes some practice with AI

The assistant can list the notes in a project, export each one in full as markdown, and read them end to end rather than sampling. It can also pull the workspace tag list, which tells it what vocabulary the team already uses for themes so the summary comes back in your language rather than its own.

What you end up with. A written pass across every note in the project, grouped into themes that use the team's existing tag names, with the notes behind each theme listed.

Start with it bringing you the information and you deciding.

Never without. Have it name how many notes it read against how many the project holds. Listing is cursor paginated, and a summary drawn from the first page reads exactly like a summary drawn from all of them.

Built on: List Notes / Export Note / List Tags / Get Project

Getting an interview transcript into the right project without the copy and paste Assumes some practice with AI

Dovetail can pull a file straight from a URL as a new note, including PDF, video and audio, so a recording or a transcript that already sits in your storage does not have to be moved by hand. The assistant can also create a note from text it has in front of it. Either way it needs the project first, and the file has to be reachable at a public URL for the import route.

What you end up with. The transcript or recording filed as a note in the project it belongs to, titled and ready to highlight.

Start with it doing the work and checking with you.

Never without. Confirm the destination project by name before the import runs, and check what the URL you are handing over exposes. A public file URL is readable by anyone who has it for as long as it stays live, and participant recordings are the last thing you want sitting behind an open link.

Built on: Import Note File / Create Note / List Projects / Get File

Writing up the synthesis so it stops living in my head Assumes some practice with AI

An insight is Dovetail's record of a finding, and the assistant can create one, attach it to a project, and revise its title and fields later as the evidence firms up. Body content is stored but not echoed back in the response, so the assistant cannot confirm what it wrote by reading its own result. Have it read the insight back afterwards.

What you end up with. A saved insight in the project, with the finding written out and the highlights it rests on cited inside it.

Start with it doing the work and checking with you.

Never without. Read the finished insight yourself before anyone else does. The assistant is generalizing from quotes to a claim, and an insight is the artifact other people will cite without going back to the raw notes.

Built on: Create Insight / List Highlights / Update Insight / Get Insight

Setting up a project for a new study Assumes some practice with AI

The assistant can create the project, create the feedback channels that will collect each source of incoming data such as reviews or support tickets, add the topics inside them, and drop in a research plan doc, all in one pass. That turns the setup ceremony at the start of a study into one instruction.

What you end up with. A project that already has its channels, topics and plan doc in place, so the first interview can go straight in.

Start with it doing the work and checking with you.

Never without. Agree the naming before it runs. This creates several linked objects at once and the names become the workspace's structure, which is tedious to unpick once other people have started filing work under them.

Built on: Create Project / Create Channel / Create Topic / Create Doc

Keeping track of who we have spoken to Assumes some practice with AI

Contacts are readable and writable, so the assistant can look up a participant, fill in the fields that were left blank after a session, and keep the custom fields such as segment or recruitment source current. It works one contact at a time against a confirmed ID rather than matching on name.

What you end up with. A participant record that reflects the session that just happened, rather than what was known when they were recruited.

Start with it doing the work and checking with you.

Never without. Have it show you the existing record before it writes. An update replaces the fields you name, so an assistant working from a partial brief can overwrite a segment or a consent field that somebody set deliberately.

Built on: Get Contact / List Contacts / Create Contact / Update Contact

Clearing out a study that is finished Assumes you have done this kind of handover before

Notes, docs and insights can all be deleted, and Dovetail's own description is precise about what that means: the item moves to the project trash and is restorable for thirty days. After that window it is gone. The assistant can export the content first so the write-up survives the cleanup even if nobody notices in time.

What you end up with. An archived copy of everything worth keeping, held outside Dovetail, and a project cleared of the material you agreed to remove.

Start with it doing the work and checking with you.

Never without. Approve the exact items by title and export them somewhere outside Dovetail first. Deletion is restorable for thirty days and permanent after that, which is long enough for everyone to forget it happened and short enough to lose the only copy of an interview.

Built on: List Notes / Export Note / Delete Note / Delete Insight / Delete Doc

What we could not establish
  • How the connector handles a Dovetail workspace with several projects of the same name was not tested here. Most actions work from a confirmed ID rather than a title.
  • Whether magic search reaches the contents of imported video and audio files, or only their transcripts and highlights, is not stated in the connector descriptions.
  • No time saving is claimed for any candidate on this page, because none has been measured.

BUILT-IN AI Ships AI features you can turn on yourself, with no developer and no new purchase.

No event triggers. Every win in this app is invoked: you or an agent asks, an action runs.

Capabilities
READ Pull records, messages, and content out on demand.
WRITE Create and update records from an instruction.
SEARCH Query the workspace in plain language.
FILE-TRANSFER Move files in and out of the app.