Start with the course
Limit the analysis to supplied documents and keep a precise record of the files used.
Student association · University of Neuchâtel
AI,
with evidence.
Smart Nexus turns raw documents into inspectable study tools. No link to a promise: see the prompts, data and results directly.
01 · Our approach
Accelerate without giving up judgement.
Useful AI makes its reasoning inspectable. It shows what it received, what it produced and where the student still needs to decide.
Make it visible
Turn a mass of information into comparable tables, priorities and explicit criteria.
Verify before acting
Flag assumptions, missing data and conclusions that require human validation.
02 · Work archive
The content, directly.
These previews come from the local Smart Nexus project files. Switch tabs to inspect four forms of AI-assisted work.
Prioritise revision by points.
The map connects formal-methods course blocks to their estimated contribution in mock exams. It separates directly assessed concepts from material that mainly provides context.
formal_methods_exam_map.xlsx · Course sheet| Concept | Contribution | Priority | Why it matters |
|---|---|---|---|
| Propositional logic | 22.5% | HIGH | Largest scored block in the logic section |
| NP-complete problems | 20% | HIGH | Applied reductions and proofs |
| Hoare triples | 15% | HIGH | Tested in both mock exams |
| Turing machines | 12.5% | HIGH | Decidability and language classification |
| Introduction to complexity | 5% | MEDIUM | Definitions and conceptual framing |
Compare semesters, not just courses.
Five portfolios combine workload, collaboration, schedule risks and complementarity. A2026, archive and “to confirm” markers preserve the level of evidence.
course_selection_A2026.xlsx · Portfolio Comparison| Scenario | Fit /100 | Workload | Main advantage |
|---|---|---|---|
| Best overall fit | 86 | 3.7/5 | Law, users, AI systems and distributed systems |
| Social / Sustainable | 83 | 3.2/5 | Interaction and public-interest topics |
| Builder / Entrepreneur | 82 | 4.4/5 | Prototype, demo, coaching and teams |
| Technical Growth | 77 | 4.0/5 | Systems, data and team software |
| AI + Decision Making | 73 | 4.1/5 | Models, data and optimisation |
Connect a course to a real product.
The UCD synthesis applies human factors, accessibility and experimentation to the ErgonomIA adaptive keyboard. Every direction leads to a concrete test.
synthese_cours_ergonomia.xlsx · Ergonomia ideas| Direction | Concepts used | Concrete test |
|---|---|---|
| Adaptive touch keyboard | Fitts, cognitive load, accessibility | Compare sizes and spacing with ≥ 6 participants |
| Configuration workflow | Wireflows, affordance, consistency | Test two Edit / Use journeys |
| Key activation | Variables, errors, tolerance | Compare press, release and dwell |
| Multimodal BLE feedback | Attention, perception, accessibility | Test visual, audio and haptic states |
Inspect the original request.
The raw prompt reveals the initial intent: maximise exam points while reducing revision time. An imperfect request can be restructured without hiding its origin.
prompt_exam_analysis.txt · local sourceOBJECTIVE Maximise the exam result and minimise revision time. REQUESTED ANALYSIS Review past exams to understand which questions earn points. Estimate the marginal benefit of exercise sheets based on their likely contribution to success. CONTEXT TO CONNECT Identify the concepts, chapters, examples, theorems, mechanisms and formulas connected to both mock exams. FIRST OUTPUT Summarise the mock exams proportionally by points.
03 · Before generating
Context is the control layer.
A good request defines permitted sources, the expected result and how uncertainty should be handled.
FIRST RULE
Ask the course lead before using AI.
Permitted uses depend on the course, assessment and citation rules. Transparency is part of the work.
01 / PERMISSIONIs AI permitted for this task?
02 / SOURCESWhich documents may inform the analysis?
03 / TRACEMust prompts and results be disclosed?
04 / VALIDATIONWhat knowledge must I demonstrate alone?
04 · The association
Learn together. Build for real.
Workshops + guests
Methods you can test.
Hands-on workshops, shared experience and outside perspectives to compare tools with their real uses.
Join the WhatsApp group ↗Open project
Explore. Test. Contribute.
Smart Nexus is built in the open. You are warmly invited to explore the project, suggest an improvement or contribute directly on GitHub.
View and contribute on GitHub ↗