Workflows in Studio

Bring a question. Move forward with an AI team.

Explore the planned AI team workflow: frame a question, compare directions and refine a proposal. Recording and reading tools already implemented in development builds help you prepare the records and evidence for your work.

ONE DIRECTION. COORDINATED WORK.

You set the goal. The team carries it forward.

The planned experience connects two loops: you and a coordinator keep the direction clear, while specialists work on focused tasks, check results and revise. Their work returns as one result you can inspect.

Illustrative coordination concept · In development. Roles and integrations may change.
OUTER LOOP · DIRECTION & DECISIONSPlanned experience
YOU · THE SHARED GOAL

Make the first step of a research tool easier.

Use the selected interviews and research brief. Ask before expanding access or changing the scope.

  1. 01Set the goal
  2. 02Shape the brief
  3. 03Review the evidence
  4. 04Choose what follows

INNER LOOP · FOCUSED EXECUTION

Brief, work, check, refine.

Shared context · Separate assignments
COORDINATOR · KEEPS THE BRIEF CONNECTED

Turn the question into bounded assignments.

The coordinator brings together the goal, selected context, ownership and what a useful result must show.

  1. MMiraResearch

    Find which part of getting started caused trouble in the interviews.

    ↳ Evidence notes + source passagesTask defined
  2. TTheoDesign

    Draft a guided first question within the agreed scope.

    ↳ A first-step proposalTask defined
  3. QQuinnIndependent review

    Check the proposal against the brief, sources and open questions.

    ↳ Findings + revision requestsReview criteria set

A check can send work back. Clarify missing context before starting dependent work.

ONE RESULT TO INSPECT

A brief the team can work from

The same goal, with clear responsibilities, limits and checks.

COORDINATOR · KEEPS THE BRIEF CONNECTED

Let independent tasks move together.

Research and design can progress separately. Each specialist keeps its own work visible; dependent checks wait for the material they need.

  1. MMiraResearch

    Find which part of getting started caused trouble in the interviews.

    ↳ Evidence notes + source passagesReading selected sources
  2. TTheoDesign

    Draft a guided first question within the agreed scope.

    ↳ A first-step proposalDrafting the approach
  3. QQuinnIndependent review

    Check the proposal against the brief, sources and open questions.

    ↳ Findings + revision requestsWaiting for the draft

A check can send work back. Only independent tasks run in parallel; dependencies remain visible.

ONE RESULT TO INSPECT

Evidence notes + a proposed first step

The coordinator brings the separate findings back to the same brief.

COORDINATOR · KEEPS THE BRIEF CONNECTED

Return a gap to the person who can fix it.

The reviewer finds that the draft assumes more interviews than the selected sources support. The coordinator sends that part back for revision.

  1. MMiraResearch

    Find which part of getting started caused trouble in the interviews.

    ↳ Evidence notes + source passagesSource passages ready
  2. TTheoDesign

    Draft a guided first question within the agreed scope.

    ↳ A first-step proposalRevision requested
  3. QQuinnIndependent review

    Check the proposal against the brief, sources and open questions.

    ↳ Findings + revision requestsGap identified

A check can send work back. Keep the failed check and the next revision connected.

ONE RESULT TO INSPECT

A proposal with a visible correction

The recommendation narrows; the uncertainty stays in the result instead of disappearing.

COORDINATOR · KEEPS THE BRIEF CONNECTED

Hand back the work and the evidence together.

The coordinator combines the revised proposal, source passages, checks and open questions into one result for you to review.

  1. MMiraResearch

    Find which part of getting started caused trouble in the interviews.

    ↳ Evidence notes + source passagesEvidence linked
  2. TTheoDesign

    Draft a guided first question within the agreed scope.

    ↳ A first-step proposalRevision included
  3. QQuinnIndependent review

    Check the proposal against the brief, sources and open questions.

    ↳ Findings + revision requestsChecks recorded

A check can send work back. Passing checks and unresolved questions are reported separately.

ONE RESULT TO INSPECT

A clearer first step · proposal v2

A guided opening question, the interview evidence behind it, and a next step to test with new users.

Back to your direction. Inspect the sources, request changes or choose the next step. A finished task does not make the decision for you.

See the team plans
Planned

Explore an idea before committing to it.

Use different roles to compare options and identify what you still need to learn.

Start with

  • A question or early idea
  • Your notes and selected sources
  • Constraints such as time, budget or audience

How the work proceeds

  1. Frame the question

    Set out the decision and its constraints. Choose members with relevant roles, such as research, design and review.

  2. Compare the options

    Discuss possible approaches with the members. Ask each to explain its reasoning, tradeoffs and assumptions.

  3. Choose what to investigate

    Challenge weak assumptions and add your own feedback. Keep the promising directions and the questions to answer next.

Example result

How should new users get started?

  • Option A: a guided setup for a first recording.
  • Option B: a sample workspace to explore first.
  • Next check: observe where first-time users hesitate.

A fictional example to illustrate the workflow.

Planned

Turn a chosen direction into a proposal you can review.

Build on selected evidence and decisions, then refine a specific version.

Start with

  • A direction you want to pursue
  • Selected sources and earlier decisions
  • The audience and requirements for the proposal

How the work proceeds

  1. Set the brief

    Explain the goal, audience and requirements. Select the material and decisions the team should use.

  2. Draft and review

    Ask a research member to organize the evidence and a design member to draft the proposal. Ask a reviewer to identify gaps and unresolved assumptions.

  3. Revise a specific version

    Give feedback and compare revisions. Accept a particular version when it meets your needs, or keep refining it.

Example result

A proposal for a small product pilot

  • Scope: test one workflow with a small participant group.
  • Evidence: findings from interviews and the product review.
  • Before approval: confirm the owner, timeline and evaluation criteria.

A fictional example to illustrate the workflow.

Implemented in development builds

Leave a meeting with a checked record.

Find the decisions and follow-up work without relying on memory.

Start with

  • A recorded meeting or imported audio
  • The questions you need to resolve
  • A meeting-summary template

How the work proceeds

  1. Capture the discussion

    Record the meeting or import its audio. Keep notes and time markers for points you want to revisit.

  2. Organize and check

    Generate a summary with a template. Check important decisions, names and tasks against the transcript and recording.

  3. Resolve a follow-up question

    Ask about the meeting. Open the answer’s numbered sources to revisit the cited transcript passage and audio, then edit a follow-up draft.

Example result

A product-review meeting

  • Decision: run a small pilot before a wider launch.
  • Follow-up: Maya will confirm the pilot participants.
  • Open question: the launch date remains undecided.

A fictional example to illustrate the workflow.

Availability and scope Recording, import, summaries and meeting questions are implemented in development builds; no public installer is available yet. Check important summaries and tasks against the original record. Ask answers support source navigation. AI team integration is planned.

See the recording steps Check current features and limits
Implemented in development builds

Read in two languages. Check the source.

Keep the original, translation, study notes and document questions together.

Start with

  • A PDF, Word (.docx), Markdown or text file
  • A passage or question you want to understand
  • Your notes and the evidence you want to keep

How the work proceeds

  1. Read with the original in view

    Open a document and switch to Side by side. Translate into Traditional Chinese for Taiwan. For a PDF, translate the current page or the document while retaining the source page layout.

  2. Clarify and check

    Generate Study notes for section key points, or ask about this document or all documents. Open cited pages and passages to check what an answer relies on.

  3. Keep the useful evidence

    Highlight passages and quote them into Markdown notes with source links. Keep your own interpretation and unanswered questions alongside the material.

Example result

Reading a workplace-learning paper

  • Read the method section beside its Traditional Chinese translation.
  • Check the participant group and study period in an Ask citation.
  • Keep the key passage and a question for your next reading session.

A fictional example to illustrate the workflow.

Availability and scope Reading, translation, highlights, source-linked notes, Study notes and document Ask are implemented in development builds; no public installer is available yet. Translation currently targets Traditional Chinese for Taiwan and, like Study notes, needs a built-in model. Ask uses your selected provider. Scanned PDFs support reading and notes; translation, highlights and Ask need extractable text. Multi-agent research comparison remains planned.

Start with the reading guide See translation modes and limits Check what is implemented

AI for paper reading

Work through a paper, with the original always close.

Use AI to translate a difficult page, ask a focused question, and organize study notes. Check the passage behind each answer and keep a useful quotation. Follow this short example to see how the reading tools connect.

Guided example · fictional paper

The paper, figures and prepared answers below are fictional. They illustrate reading tools implemented in development builds.

Choose a step. Open a numbered source to check the original passage.

A difficult page becomes easier to follow.

Keep the English page and its Traditional Chinese translation together. The matching headings help you compare a method or result without losing your place.

Documentsshort-review-sessions.pdfSide by sideTranslate this page
OriginalTraditional Chinese translation
OriginalPage 1

Short review sessions and recall

A fictional classroom study for this guided example

Method

Twenty-four adult volunteers were randomly assigned to two groups of 12. Both groups read the same 800-word explanation over a two-week period.

Review task

One group reviewed the material in three five-minute sessions. The other reread it in one 15-minute session. Both took the same ten-question quiz seven days later.

Findings

The spaced-review group averaged 7.8 correct answers out of 10; the rereading group averaged 6.2. This difference describes the participants in this example.

Limitations

The sample was small and included only adult volunteers. One short quiz cannot establish effects for other populations or longer periods.

Average correct answers · fictional data
Spaced review7.8 / 10
Rereading6.2 / 10
TranslationPage 1

短時間複習與記憶表現

供導覽示例使用的虛構課堂研究

研究方法

24 位成年志願者被隨機分為兩組，每組 12 人。兩組在兩週期間閱讀相同的 800 字說明。

複習任務

一組分三次複習，每次五分鐘；另一組集中重讀一次，共 15 分鐘。七天後，兩組接受相同的十題測驗。

研究結果

分次複習組平均答對 7.8 題，集中重讀組平均答對 6.2 題，滿分為 10 題。此差異只描述這個示例中的參與者。

研究限制

樣本規模小，且僅包含成年志願者。一次簡短測驗無法確認其他族群或更長時間的效果。

平均答對題數 · 虛構資料
分次複習7.8 / 10
集中重讀6.2 / 10

See the original and translated PDF page together.

A focused question, with a passage you can inspect.

Ask about a claim, method or limitation. Then open the answer’s source and decide whether the original passage provides enough evidence.

Documentsshort-review-sessions.pdfAsk

Ask about This document

Prepared example question

Does this study show that spaced review works for everyone?

Prepared example answer

No. The example studied only 24 adult volunteers and measured one short quiz. Its result suggests a direction to investigate, but it does not establish effects for everyone or over longer periods.

Sources1 · Method and limitations

Select [1] in the prepared answer to open the original.

Study notes give the next reading pass a structure.

Review section points and follow their source links. The example keeps the sample, result and limit separate so a useful result does not become an oversized claim.

Documentsshort-review-sessions.pdfStudy notes
Study notesWrite study notes

  1. A small, defined sample

    24 adult volunteers; two groups of 12.

    1 · Method

  2. A result within this example

    Average correct answers: 7.8 for spaced review, 6.2 for rereading.

    2 · Findings

  3. A boundary on the conclusion

    The sample and single quiz do not establish broader or long-term effects.

    3 · Limitations

Check against the original

Open a source link beside a point to check it.

Carry a useful passage into your own notes.

Quote a passage with its source link, then add your interpretation or a question to investigate. You can return from the note to the original page.

Documentsshort-review-sessions.pdfNotes
OriginalPDF · 1

Short review sessions and recall

Method

Twenty-four adult volunteers were randomly assigned to two groups of 12. Both groups read the same 800-word explanation over a two-week period.

Choose “Quote in notes” to keep the selected passage together with its source link.

NotesSaved in this example

What this small study can tell us

Twenty-four adult volunteers were randomly assigned to two groups of 12. Both groups read the same 800-word explanation over a two-week period.
Source: short-review-sessions.pdf · page 1 · Method

My reading question: would the same pattern hold with a larger, more varied group?

Use “Quote in notes”, then follow the saved source link.

Before you start

Translation and Study notes use a built-in local model; translation currently targets Traditional Chinese. Ask uses your selected provider and may send your question and relevant excerpts to a remote service. PDF translation may shrink or clip text; check the original. These tools are in development builds; a public installer is not available yet.

Read PDF · Word (.docx) · Markdown · textOpen the reading guide See current features and limits

What would you like to read, record or explore?

Tell us about your reading or meeting workflow, or ask about AI team development.

Email us