QualiSense — Making human meaning

Version

Making human meaning · Researcher-led qualitative analysis
Local workspaceEncrypted locally after setup · Back up regularly
QualiSense · Free Web Edition

Start by creating a project, adding transcripts under Manage Transcripts, and coding in Transcript Coding. Your data is saved in this browser, not on a central server. Download an encrypted backup before changing devices or browsers.

Optional local AI setup

Enable the same Qwen2.5 3B model for AI Coding Suggestions and the QualiSense Research Assistant. The first activation can require a substantial download and a WebGPU-capable browser. The model may be reused from your browser cache on later visits, but needs to be loaded into memory again. Manual research features work without AI.

AI not activated

Model download and runtime files are retrieved from the configured model provider. Transcript contents are not uploaded by this setup step. Activation is optional.

Projects

Projects are isolated: each has its own transcripts, codes, themes, memos and suggestions. Use the selector to switch.

Research workspace

Your selection documents your approach; it does not automatically impose a coding framework or theoretical interpretation.

Research question & conceptual definitions

Optional researcher-authored context for offline coding suggestions, especially when starting with an empty codebook. Add up to three concepts and their working definitions. These are sensitising concepts, not predetermined codes; the researcher approves every proposed code.

This describes the person providing interview data, not their occupation. The AI must not call a participant a consultant, manager or leader unless the source explicitly establishes that role. This setting applies to new AI coding suggestions; it does not rewrite transcripts or past suggestions.

This workspace keeps descriptive evidence, researcher coding and interpretative hypotheses separate. No transcript is sent to an external AI service.

0
Transcripts
0
Codes
0
Coded extracts
0
Themes

Suggested research workflow

Import anonymised transcripts → select verbatim passages → assign researcher-approved codes → write reflexive memos → develop candidate themes → compare evidence → export a transparent analysis record.

Privacy: data is stored in this browser's local storage, not encrypted. Use anonymised material only, back up regularly and avoid shared devices. This prototype is not an institutional data-security solution.

Transcript Coding

One workspace for reading interviews, selecting exact verbatim evidence, assigning existing or new codes, and writing analytic memos. AI Suggestions remains a separate optional stage.

Highlight text in the centre and click + New code to create and assign it immediately. For existing codes, click a code in the left panel or use the floating toolbar. Click an existing highlight to inspect its coding. All edits retain the source quotation and character positions.

Select a transcript

No transcript selected.
Select text in the transcript to code it.
Classic passage coding controls

Select a transcript

Highlight an exact passage below, then click “Capture selection”. The selected quote is saved verbatim with its location in the transcript.

No transcript selected.
No passage captured.

Code captured passage

Extracts in selected transcript

Codebook — review and refine

Create codes and write their definitions directly in the Codebook, without needing to select a transcript passage. Here you can also review and edit code definitions and memos, inspect assigned quotations, reassign or remove evidence, and manage code hierarchy, merging and deletion.

Codebook guidance

Expand a code to inspect all assigned verbatim evidence. Edit its definition in the box beneath the code and click Save definition. Use Edit to update the name, memo and analytic level.

New codes are saved to the active project and can be assigned to verbatim evidence later.

Add a new code

Create a code and definition independently of transcript coding.

Codes and definitions

Edit names, definitions, levels and hierarchy without detaching existing quotations. Reassign or merge codes when needed.

Merge codes

Choose the code to merge from and the code to retain into. Existing coded passages will be preserved.

Coding cycles are researcher-defined. Existing codes are first cycle by default; you can change any code below without losing its evidence.

Manage Transcripts

Upload, review and delete interview transcripts in the active project. Coding takes place separately in Transcript Coding. Deleting a transcript also deletes its coding assignments and pending suggestions; export a backup first.

Transcripts

Enter a transcript name before choosing a file. The name you enter will be used throughout QualiSense.

Cross-case verbatim comparison

Select a code to compare its verbatim evidence across transcripts. Quotations are displayed exactly as coded; empty cases are shown explicitly. This is a descriptive view, not an automated interpretation.

Evidence Matrix

Compare coded evidence across participants. Select any cell to review quotations, remove an incorrect coding assignment, reassign evidence to another existing code, or create a new code from an extract. The original transcript is never modified.

Cross-participant evidence matrix

Counts of coded extracts by transcript and code. Click a cell to inspect the supporting quotations.

Select a matrix cell to review evidence.

1. Conceptual Clustering Workspace — unlock your workspace to begin.

2. Theme Development Studio — unlock your workspace to begin.

3. My Themes — Review & Manage

Review saved themes, their linked codes and verbatim evidence. Themes saved from the Theme Development Studio appear here. Drafts that have not been added to the Themes list remain in Section 2.

4. Visual Theme Designer

Build a researcher-controlled diagram of themes, codes and their relationships. Export your diagram for publications or presentations.

Analytic and reflexive memos

Memo journal

Local AI coding assistant — Qwen2.5 3B

A small language model runs on this device. The first use downloads model files from a public model host; no transcript, codebook or research question is sent to an AI API. Chrome or Edge with WebGPU recommended. Model output may be unreliable; review every suggestion.

One on-device AI model for optional coding suggestions. You review all suggestions before applying them.

The local AI engine is managed on the Overview page. Load it there to enable coding suggestions and broader Research Assistant questions.

First- and second-cycle coding suggest ONLY existing codes in the selected cycle. Exploratory coding alone proposes new codes. All suggestions require researcher approval.

If no existing code fits, the AI leaves the passage uncoded rather than inventing a code. New code creation is available only in Exploratory coding.

Suggestions are never applied automatically. Review the verified evidence, then explicitly choose Accept and apply, Save to Codebook only, or Discard.

Previously queued suggestions (legacy) — open only to finish earlier reviews

Previously queued suggestions

Open a suggested passage below. Highlight the exact words, choose an existing or new code, then assign it. Use Discard suggestion to remove an irrelevant suggestion individually.

Export your research

JSON preserves projects, transcripts, coded selections, codebook, themes, memos, pending suggestions and audit events. CSV files are compatible with Excel and other research workflows.

đź“‚ Import or restore QualiSense projects

Choose an encrypted backup, older unencrypted backup, or compatible project JSON. QualiSense detects the format and previews its projects before making changes.

Choose a JSON file to preview its contents. Nothing will be changed until you confirm.

Adding preserves your existing projects. Replacing overwrites the current browser workspace after confirmation. Download a backup of your current work first.

Audit trail