CMOC 2
C Learners with long, costly journeys
M feel the course is within reach
O and keep attending
R when travel is paid for
Back to Oriel
Oriel
Coming October 2026
Oriel
Coming October 2026
The feature guide, workflow by workflow
From first source to final finding, without the wait.
Oriel is a fast, keyboard-friendly desktop app for qualitative research, realist programme theory, evidence synthesis, and behavioural science. Follow the ten workflows below, and try each one with fictional material.
Free, with Windows first. Every workflow states its platform limits and the choices involved in AI and collaboration.
Built to keep up
Fast enough to stay out of your way.
Analysis is thinking time. Oriel is designed so that opening, coding, searching, and moving around feel immediate, and you keep your attention on the material.
-
A Rust core on your own computer
Each project is a single local file. Opening a source, applying a code, or running a query works on your machine and never waits on a server.
-
No Save button to remember
Coding is saved the moment you apply it. Notes and descriptions save when you click away, unsaved text survives a crash, and anything can be undone.
-
Indexed and light on memory
Text search runs on an index. Long PDFs draw only the pages near you, large datasets scroll virtually, and imports stream one file at a time.
-
Keyboard-first when you want it
Jump anywhere, code by typing, and step through sources and passages without reaching for the mouse.
⌘K Go to · ⌥← Back · ⌥PgDn Next source
Shortcuts are shown for macOS. On Windows, use Ctrl for ⌘ and Alt for ⌥. Shortcuts can be rebound in Settings.
Work together
Collaborate through OneDrive, Teams, Dropbox, or your network drive.
No Oriel server and no new account: choose a shared folder your team already uses, and each colleague keeps a fast local copy that stays in step.
Workflow 01
Drop it in. Oriel sorts the rest.
Start a project, bring in documents, datasets, recordings, images, web material and references, and keep the originals connected.
Drag files or whole folders straight into the window. Oriel files them into matching folders, shows a summary, and opens the first one.
Begin in seconds
Choose a qualitative, realist, or qualitative evidence synthesis project, or open an editable fictional example with interviews, starter codes, cases, saved queries, a memo, and a programme theory.
Drag in files and folders
Drop files on a source folder, the Sources list, or anywhere in the window. A dropped folder becomes a source folder, with its sub-folders inside. Everything is also under File → Import.
Read documents in a useful form
Import TXT, Markdown, DOCX, DOC, RTF, and PDF files. A Word layout preview sits alongside the stable text used for coding. DOC and RTF formatting is retained on macOS; Windows imports their text only.
Work with datasets, recordings and images
Import XLS, XLSX, CSV, and TSV files, choose columns to code, and create cases from rows. Bring in audio, video, attached transcripts, and images.
Capture online material
Import saved HTML pages, X (Twitter) archives, and Mastodon exports, or capture a web page by its address. Online capture needs an internet connection.
Keep references with their documents
Import RIS, BibTeX, or EndNote XML, paste a reference list, or look up a DOI. The Bibliography, under Outputs, shows every reference once, with the documents it belongs to, in APA 7, Harvard, Vancouver, or Chicago author–date style.
Find and track every source
Search sources by name or folder. The Source overview table filters by folder, type and status, adds columns such as file size or linked memos, and opens a source’s coding references in one click. Tick several sources to act on them together.
Bring an NVivo project across
Import Windows .nvp or Mac .nvpx files without NVivo installed. A REFI-QDA .qdpx export or a codebook, preferably .qdc, restores names; without one, names may remain placeholders.
Interview – Ben.docx
Focus group 2.m4a
Survey responses.xlsx
Sources
- Interview – Amira
Drop files here
Workflow 02
Stay with the passage. Code as you read.
Code text, PDF and image areas, recordings, transcripts, and dataset cells, and move around a large project without losing your place.
Select words and start typing: the Quick codes card filters as you type, and Enter applies the code. Press ⌘K and type three letters to open anything.
Quick codes beside your selection
A small card appears where the selection ends. Type to search the whole hierarchy, browse into a parent code, or create a new code and apply it in the same step.
Code more than ordinary text
Code document text, PDF or image areas, recording time ranges, transcript passages, and dataset cells, with in vivo codes named after a participant’s own words.
Go to anything
Press ⌘K to find sources, codes, cases, memos, findings, studies, saved queries, sets, settings, and commands by name. Recent choices come first.
Back, Forward, and next passage
Go back to where you were reading, as in a web browser, across the last 50 places. Step to the next or previous source, or through each coded passage in turn.
Track coding progress
Each source shows Not started, In progress, or Coded. Oriel works out the first two from the coding; only a person can mark a source Coded.
Maintain a coherent codebook
Nest, reorder, recolour, describe, and merge codes. Fold the Codes panel to a slim rail when you want more room to read.
Annotate and highlight
Add annotations, and highlight passages in six colours, separately from coding. Choose whose coding is visible and show coding stripes beside a source.
Return to excerpts
Group and filter excerpts by code or source, search within them, and open the original context when a quotation needs closer attention.
- Barriers
- Barriers › Cost of travel
- Barriers we expected
Sources click a ring
Type in the box, use ↑ ↓, then press Enter.
Workflow 03
From recording to codable text, on your computer.
Transcribe interviews, label speakers, and make scanned PDFs searchable, without uploading the material for processing.
Transcribe a whole folder of recordings in one go, and jump from any transcript line straight to that moment in the recording.
Transcribe locally
Download a Whisper speech model (Base, Small, or Large Turbo), then transcribe one recording or a batch. Recordings of up to six hours in common audio and video formats are supported.
Check speakers and wording
Identify speakers automatically or give the number of speakers, check the transcript against playback, and follow linked passages back to the recording.
Rename speakers to pseudonyms
Rename speakers in one undoable step, for transcripts you made or attached. Giving two speakers the same name merges them.
Recognise printed text in PDFs
Run text recognition on scanned PDF pages so their words can be selected, searched, and coded. It uses the built-in macOS or Windows engine and its languages; it is not available on Linux.
- Speaker 1What made it easier to keep coming back?
- Speaker 2Honestly, the bus pass. Without that I couldn’t.
- Speaker 3And knowing someone would be there who knew my name.
- Speaker 1So the people mattered as much as the course?
Platform notes: Windows transcription needs the graphics runtime supplied by a compatible driver, even when using CPU fallback, so virtual or remote desktops may lack it. Local processing concerns the transcription or recognition step; if you choose shared-folder collaboration, your folder provider also copies the project’s recordings and transcripts.
Workflow 04 · Optional
Suggestions you inspect, never decisions it makes.
Local AI runs on your own computer. It can suggest coding, draft code descriptions, and help draft CMOCs, and nothing is applied until you accept it.
Review suggestions in a single list: click a quotation to see it in its source, then accept or dismiss it. Accept all shown and Dismiss all shown clear the list in one go.
Check the computer and model
Oriel checks your hardware, downloads a helper program and model, and offers a speed test on a fictional task. Model size, memory, and graphics hardware affect speed and quality.
Review coding in context
Choose sources and existing codes, leave out interviewers or facilitators, and follow progress part by part. Each suggestion shows its passage, code, rationale, and confidence.
Keep an AI run distinguishable
Deliberately save a run as a separate AI coder instead, so it can be compared with your coding and removed as a whole.
Explore proposed new codes
An Experimental option proposes codes for material your codebook may not cover, and local AI can draft code descriptions.
Draft or revise CMOCs
Request a CMOC from selected passages, codes, or claims, or a revision informed by linked evidence, and review each draft.
Record what assistance was used
Inspect methods details for the model, prompt, and approach. The model is unloaded when Oriel closes and is never left running.
Honestly, the bus pass. Without that I couldn’t.
Barriers › Cost of travel High confidence
Travel costs are named as the condition for attending.
Knowing someone would be there who knew my name.
Peer support Medium confidence
Recognition by others is described as a reason to return.
The room was on the second floor, no lift.
Trust Low confidence
Possible link to confidence in the venue.
3 suggestions to review
The hardware trade-off: keeping AI processing local needs powerful hardware for practical performance, particularly for larger models. Current support is Apple silicon Macs with macOS 13.3 or later, or Windows PCs with a graphics processor and an up-to-date driver. Local AI is not available on Linux yet. Manual coding and analysis remain fully available without AI.
Workflow 05
Move from a question to the passages behind it.
Organise cases and attributes, then search, count, compare and chart, always one click from the underlying material.
Click any matrix cell, network link, or chart point to read its passages. Save a query once and re-run it as the coding grows.
Cases, classifications and sets
Represent participants, organisations, or settings as cases, give them typed attributes in an editable sheet, and group items into reusable sets.
Find words and coding
Search text exactly or by word stem with keyword in context, explore word frequencies, and combine codes, cases and attributes with AND, OR, NOT, NEAR, and PRECEDING.
Compare in a coding matrix
Cross codes with cases, attribute values, or studies in a heatmap with totals, inspect each cell, and export the table.
Write a framework matrix
Build a case-by-code matrix of your own written summaries, with the supporting evidence close at hand.
See relationships and hierarchies
Explore code co-occurrence as a table or network, and compare coding across the hierarchy as a treemap or sunburst.
Explore saturation
See when new codes, claims, or CMOCs stop appearing across ordered sources, with the Guest, Namey and Chen (2020) and Francis et al. (2010) stopping rules.
| Code | Learners | Tutors | Volunteers |
|---|---|---|---|
| Cost of travel | |||
| Peer support | |||
| Trust |
Click a cell to read its passages.
Workflow 06
Test the explanation without losing the original.
Build realist programme theory from an initial explanation to evidence, testing, and refinement, and connect it to behaviour change techniques and mechanisms of action.
Collect evidence straight from a CMOC page. Producing a refined theory is a single step that you can undo.
Start from an existing explanation
Choose realist evaluation or review, write an initial programme theory, import one from a table, or build it from codes. Use CMO or ICAMO, with optional separate Resource and Reasoning fields.
Link each part to its material
Connect passages to the whole CMOC or to a component, with a role, a stance, and a note. Rate source rigour and relevance separately from the rigour of the whole CMOC.
Run a testing round
Lock the original wording, then confirm, refute, refine, split, merge, or set aside each CMOC with reasons and evidence. Add emergent CMOCs found in the data.
Trace theory development
Work in a diagram, table, or testing board. Save versions, compare them, and export a theory development log.
Match techniques and mechanisms
Link behaviour change techniques and mechanisms of action to CMOC components, with their definitions and identifiers, alongside your own wording.
Keep the vocabulary traceable
Pin the project to an ontology release, review changes before updating, and include linked terms in CMOC tables and reports.
Workflow 07 · New
From a stack of papers to findings you can trace.
Qualitative evidence synthesis follows one pathway: link reports to studies, agree a protocol, extract and appraise, reconcile, then use the evidence in your findings.
Create a study from every paper in one step. Fields save as you type, and Agree all matching values settles every identical answer at once.
- 1Link reports to studies
- 2Agree the protocol
- 3Extract and appraise
- 4Reconcile
- 5Use in findings
Keep studies and reports distinct
A study with several papers counts once, and a paper describing several studies links to each. Create studies from sources in bulk, merge them, record sample overlap, and import screening outcomes from CSV or Excel. Studies are never matched by title.
See progress at a glance
The Studies list doubles as a progress table: reports linked, your extraction progress, reviewers who have submitted, agreed values, and appraisal status.
Agree one review protocol
Build extraction and appraisal templates from editable starters, with text, number, date, choice, and repeatable group fields. Used versions are frozen, and reviewers move their values to a new version with a preview.
Extract beside the report
Each field autosaves while the report stays open beside the form. Mark a field Not reported or Not applicable, and add exact supporting quotations from your selection.
Appraise without a score
Record structured appraisal, such as methodological limitations, question by question with your rationale. Oriel calculates no quality score.
Reconcile independent reviews
Once two reviewers have submitted, compare their values side by side, agree what matches in one step, and resolve the differences with a note. Later changes mark an agreement Needs review.
Optional AI extraction proposals
Local AI can propose values beside each field, with the quotations it relied on. Every proposal needs your acceptance. AI never appraises, never assesses CERQual, and never submits for you.
Analyse and report by study
Scope any query to studies, give each study its own matrix column, and group excerpts by study. A quotation added to a finding records that study’s contribution, with its codes.
Judge confidence with CERQual
Write your CERQual explanations and overall confidence. Changes to the evidence mark them Needs review. Sensitivity analysis shows a finding without a study, without saving anything.
Export the review tables
Preview and export study characteristics, agreed extraction, appraisal, contributions, lineage, and a Summary of Qualitative Findings to Word, Excel, or CSV.
| Field | Reviewer A | Reviewer B | Status |
|---|---|---|---|
| Setting | Community centres | Community centres | Matches |
| Sample size | 24 | 24 | Matches |
| Data collection | Differs | ||
| Recruitment | Not reported | Not reported | Matches |
3 match · 1 differs
Scope: Oriel imports screening outcomes but does not run searches, deduplicate records, or draw flow diagrams. Shared project files do not technically blind reviewers from each other, and models can miss values, so read the reports and complete missing fields yourself. Evidence synthesis can be switched on alongside realist tools.
Workflow 08
Claims that carry their evidence.
Think in memos and a research journal, then write findings whose quotations, cases, and negative cases stay attached all the way to export.
Add the last passage you selected to a claim in one click, and see its spread across cases update immediately.
Think in linked notes
Write memos linked to sources, codes, or other items, and keep a dated research journal with tags, links, and search.
Build sections and claims
Organise findings into sections and claims, marked Draft or Settled, and link them to quotations, codes, and CMOCs.
Review the spread of evidence
See which cases support a claim and which have no evidence, whose voices appear across the write-up, and configurable prompts for patterns that need another look.
Record negative cases thoughtfully
Read passages that might challenge a claim, then mark them as contradicting, or record why a passage does not contradict it.
Export a connected account
Export findings to Word, PDF, or Markdown, with quotation details and optional appendices: a claims-by-cases evidence table and a log of considered negative cases.
Reports and bibliography
Preview and export passage, codebook, query, matrix, and chart reports. Copy formatted references or export them as RIS.
Practical barriers can prevent people from taking part.
Spread across cases
- Amira
- Ben
- Chloe
- Dev
- Eve
Possible negative case “I’d come even if I had to walk.” · Dev
Workflow 09
Collaborate through the storage you already use.
Share a project through a folder in OneDrive, Teams or SharePoint, Dropbox, Google Drive, iCloud Drive, or a network drive. There is no Oriel server to run and no new account to approve.
Each person works on their own local copy at full speed, online or offline. Oriel sends your changes after a short pause in your work, at most about a minute later, and checks for colleagues’ changes every few seconds while the project is open.
YouLocal copy · full speed, online or offline
A shared folder you already have
- OneDrive
- Teams & SharePoint
- Dropbox
- Google Drive
- iCloud Drive
- Network drive
Your colleaguesEach with their own copy and coder name
- 1Click Collaborate and choose an empty folder in your cloud or network drive
- 2Share that folder with named colleagues through your provider
- 3Colleagues choose Join a shared project and save their own copy
Works with the storage you have
Any folder your computer syncs will do: OneDrive, Dropbox, Google Drive, iCloud Drive, or a network drive. Oriel uploads nothing itself; your provider copies the files and controls who can open them.
Teams and SharePoint channels
Add a channel folder to your computer with the OneDrive app’s Add shortcut to OneDrive, then choose it in Oriel like any other folder. The guide walks through it on Mac and Windows.
Never a live project file in the cloud
Your working copy stays outside the shared folder, and Oriel refuses to keep it inside. The folder holds only small change files that are never rewritten, so a sync client cannot corrupt an open database.
Share from the start, or later
Tick Share this project with others when you create a project, or click Collaborate on an existing one. New members join from a recent snapshot instead of replaying every change.
Continue offline, then catch up
Keep working when disconnected; pending changes are exchanged when the connection and folder are back. Collaborate shows the status, and Sync now exchanges at once.
Keep cloud folders downloaded
If syncing keeps failing, Oriel suggests the folder may not be fully downloaded, and on a Mac it counts iCloud files still waiting. The guide shows how to keep OneDrive, Dropbox, and iCloud folders on your computer so syncing stays quick and works offline.
Everyone’s work stays identifiable
Each copy codes under its own coder name, and Undo only undoes your own changes. Concurrent coding of the same passage keeps both coders’ references and attribution.
Edits made at the same time
When two people edit different parts of a memo or journal entry, both changes are kept. When they change the same field, Oriel shows the saved text and asks which to keep, and Conflicts can put back an overwritten value.
Compare coders
Use blind coding and coder groups, compare percentage agreement and Cohen’s kappa, and open each disagreement in its source. A separate AI coder can be compared in the same way.
Advanced: private coding service
Turn this on in Settings → Advanced if your team runs one. It needs an administrator and a self-hosted service, shares coding only, and excludes source text, code names, memos, annotations, and attributes.
You
Early impressions
Travel keeps coming up.
Sam
Early impressions
Travel keeps coming up.
Sam saved a different title while you were typing.
Who can see what: a shared folder carries the whole project, including original recordings, transcripts, and deleted text kept in the history. Oriel adds no encryption of its own and cannot remove a person, so use a provider your institution and ethics approval allow, share with named people only, never by public link, and remove access at the provider when someone leaves. Cross-platform shared-folder collaboration has been tested through iCloud Drive and local network folders; OneDrive, Dropbox, and other providers, and individual failure scenarios, remain unconfirmed. Independent security review and some private-service work remain outstanding.
Workflow 10
Recover easily. Exchange openly.
A complete history, automatic backups, and routes into and out of other software, so your work is never stuck.
Undo your own changes step by step, and restore a backup in a few clicks. Backups store each source file once, so they stay small.
History and recovery
Undo and redo, item history, checkpoints, earlier versions, and restoring an item as it was. Unsaved text is recovered after a crash.
Automatic backups
Every 10 minutes while you work, by default, keeping the newest 10 plus one a day for a week, in a folder Oriel looks after.
Exchange through REFI-QDA
Import and export REFI-QDA projects and codebooks. Formats organise projects differently, so some information is lost, and export back to NVivo is not fully supported.
Work comfortably
Choose themes, text sizes, reduced motion, and rebindable shortcuts, and carry your preferences to another computer with a settings profile.
NVivo attribute values, memo text, and saved searches do not transfer on import. REFI-QDA import does not keep coding authorship or dates, and saved queries, programme theories, evidence links, and findings are not exported. Local project and backup files are not encrypted or password-protected by Oriel.
No features match. Try a shorter word.
See the workspace in action.
Try the Quick codes and realist theory previews with fictional material, or register your interest in the October 2026 release.