FREE & OPEN SOURCE · LOCAL AI

Your materials.Your AI.Your exam preparation.

Turn your lecture PDFs and Markdown notes into source-grounded questions. Explain what you know, check the original source, and practise for oral exams with AI running on your computer.

No account required · Local AI · Open source

Eight ways to practise your course material
Colloquily practice picker with eight illustrated modes, exam navigation and the version footer.
Actual app UI · English · Synthetic course workspace, not measured AI output

FROM LECTURE TO EXAM

A little structure. A lot more practice.

Start with the materials you already have. Build a study routine around what you need to learn.

01

Add your materials

Create an exam and import lecture PDFs or Markdown notes. Choose Prepare my practice to prepare eligible source evidence and questions.

02

Explain from memory

Choose a practice mode, answer in your own words and submit your written explanation for local AI feedback.

03

Check and revisit

Open the cited PDF page, review missing points and return to the questions that need more work.

CHOOSE YOUR PRACTICE

One concept or the full colloquium.

All eight modes include a guide to the learning goal, how to practise and what the session does.

Single question

Explain one concept, its reasoning and an example.

Feedback: After your answer

Random practice

Recall the right idea across a mixed selection, with question count and topic filters.

Feedback: After the run

Weak topic

Revisit questions due for review, then the remaining pool. Without history, use the usual pool.

Feedback: After the run

Oral exam simulation

Rehearse structured explanations within a time budget, with count and filters.

Feedback: After the run

Rapid practice

Give concise answers on familiar material in a quick round.

Feedback: After each answer

Subject simulation

Sustain explanations across the question pool for one exam.

Feedback: After the run

Multiple subjects

Practise switching between two or three exams in consecutive timed blocks.

Feedback: After all blocks

Full colloquium

Rehearse exactly three exams. Start with 15 minutes per subject or adjust the duration.

Feedback: After all blocks

Deferred evaluation starts automatically when the run finishes. Preparation counts toward the timer; multi-subject durations apply to each subject. Speak aloud if useful, then submit text: voice and delivery are not assessed. An unsent draft is not submitted on timeout.

BUILT AROUND YOUR COURSE

Make your next study session count.

One place to turn reading into active exam preparation.

Source-grounded questions

Questions need current, accepted evidence and supporting quotes. Unsupported candidates are rejected instead of replaced with filler.

Eight practice modes

Start with one concept, revisit weak topics or rehearse a timed session across up to three exams.

Chat with your materials

Ask follow-up questions in a local conversation for each exam, with references to the course evidence.

Original source insights

Open the cited PDF page inside practice to check formulas, diagrams and the original context.

Useful session review

Review your submitted text, question scores and feedback-based next steps. Skipped and ungraded answers stay visible.

Local AI. Your control.

Documents, answers and evaluations stay on your computer. Each exam keeps its own materials and progress.

AI questions and feedback can be wrong. Check them against the original sources; scores are study aids, not predictions of an exam grade. Scanned PDFs need separate OCR.

FROM ANSWER TO NEXT STEP

Explain. Check the source. Try again.

Keep your own explanation at the centre, then use the source and feedback to decide what to practise next.

Focused Colloquily practice screen with a question about mass and energy and a written answer.
Actual English app UI · Synthetic answers and feedback for demonstration · Open full-size screenshot

PRIVATE BY DESIGN

Your files stay yours.

Local AI inference runs on your device through Ollama. Your lecture files do not need to be sent to a cloud AI provider. The desktop app has no cloud AI fallback, cloud account, or analytics service.

Internet access is used for installation, model downloads and update checks. This website uses optional analytics only with your consent; hosting providers process requests to serve it. GitHub and Ollama have their own privacy policies.

YOUR NEXT STUDY SESSION STARTS HERE

Ready for your next exam?

Choose your computer. Bring your lecture notes.

macOS

Apple Silicon

macOS 14 or newer

Download

Windows

x64

64-bit Windows · WebView2 runtime

Download

New in 0.1.10

The desktop sidebar now fits your window while the main page scrolls. Navigation scrolls independently in short windows, keeping the exam picker and About footer visible. Completed evaluations also return their final saved grade consistently.

Read the release notes

Version 0.1.6 and later can receive signed automatic updates. The app backs up before installation and waits for an idle Home screen before restarting. Choose Later or disable automatic updates in About. Older versions need one manual upgrade.

Preview release with signed updater payloads. Apple notarization and Windows publisher signing are separate: check the platform trust reports on the release page. Your OS may warn or block installation. Verify SHA256SUMS.txt and get the matching license bundle. Clean-machine installation and full hardware coverage remain separate checks.

Installation instructions
  1. Download the installer for your architecture and verify its checksum on GitHub.
  2. macOS: open the DMG and drag Colloquily into Applications. Windows: run the x64 installer. Follow the linked installation guide for unsigned previews.
  3. Install and start Ollama separately, then follow AI setup in Colloquily to prepare local models. Model downloads need internet access and free disk space.
View all releases

OPEN TO EVERYONE

Free to use. Open to improve.

Colloquily is open source under Apache-2.0. Have an idea for better exam preparation? Help shape what comes next.