Veritas

What it is

A closed-corpus thesis engine: it operates entirely on materials you supply (PDF articles, books, lecture slides, survey exports, interview transcripts, spreadsheets). It organizes, interprets, tests, and writes, turning your raw stack into a defensible, KEDGE-compliant thesis.

What it is not

• It does not search the web or add outside sources.

• It does not fabricate citations or data.

• It does not submit work on your behalf or promise to “evade detection.”

• It is a drafting and analysis assistant; you remain the author.

Inputs it accepts

• Literature: PDFs (peer-reviewed articles, white papers, reports), slide decks.

• Data: CSV/XLSX from surveys (e.g., Google Forms/Qualtrics), experiment logs, coding sheets, interview transcripts.

• Constraints: school guide (formatting, structure, word count), rubric, required theories, required sections.

• Style seeds: 1–3 pages of your writing to match tone (used as suggestions; you approve/overwrite).

Core capabilities

1) Corpus ingestion & organization

• Extracts metadata (title, authors, year, venue, DOI/URL) and builds a study matrix (methods, sample, key findings, limitations, tags).

• Deduplicates and clusters papers by theme, method, theory, and your custom tags (e.g., Ownable/Personalised/Immersive; trust; friction).

• Links each claim to its evidence trail (paper → page → passage), so every paragraph has sources.

2) Literature synthesis (thesis–antithesis–synthesis)

• For each theme, generates:

  • Thesis (mainstream or supportive view).

  • Antithesis (conflicting, critical, or boundary conditions).

  • Synthesis (what survives after reconciling evidence, with scope/limits).

• Produces “so-what” bullets that directly inform your hypotheses and managerial implications.

3) Hypotheses & theoretical scaffolding

• Maps your corpus to appropriate lenses (e.g., TAM/UTAUT2, privacy calculus, Uses & Gratifications, flow, social/telepresence).

• Proposes directional hypotheses grounded in the literature clusters, each traceable to sources and measurable with the data you supplied.

• Builds a conceptual pathway (constructs → mechanisms → outcomes) and an RQ–theory–metric map showing exactly how each hypothesis will be tested.

4) Method planning from your data (no new collection)

• Reads your survey instrument and outputs:

  • Variable dictionary, scale reliability checks (α/ω), index construction rules.

  • If you used Best–Worst Scaling/MaxDiff: task balance checks, part-worth estimation plan (MNL/HB), and segment definitions.

• Generates analysis scripts (R/Python, optional) for full reproducibility and a bulletproof Appendix.

5) Analysis & results writing (from your files)

• Runs the planned analyses only on your data: descriptives, correlations, regressions/ANOVA/SEM, or BWS part-worths and confidence intervals.

• Auto-creates tables/figures (publication-style), captions, and neutral, non-overclaimy narrative text.

• Adds a robustness subsection (alternative specs, sensitivity checks) where feasible.

6) Full-thesis drafting (school-specific)

• Outputs a complete draft following your programme’s structure (e.g., KEDGE: Intro → Research Question → Literature Review → Hypotheses → Model → Methodology → Results → Discussion → Contributions → Managerial Implications → Limitations → Future Research → References → Appendix).

• Inserts traceable in-text citations (Author–Year) and builds an APA 7 reference list from your PDFs—no phantom citations.

• Provides chapter roadmaps, signposting, and bridging paragraphs so the story flows.

7) Managerial implications & playbook

• Converts findings into design principles, risk/trust mitigation, pricing/packaging levers, and a phased implementation roadmap (MVP → scale).

• Includes a managerial checklist and a measurement plan (leading indicators, retention cohorts, advocacy proxies).

8) Voice alignment & transparency

• Suggests phrasing in your style (from your samples), but keeps change-tracking on: every AI-assisted sentence is reviewable.

• Flags any paragraph with thin evidence or over-extrapolation for you to revise.

9) Integrity & compliance rails

• No external retrieval: everything is drawn from your uploaded corpus.

• Citation integrity: every cite is grounded in a specific PDF and page span.

• Originality support: paraphrase suggestions and quote extraction with page refs; you decide what to keep.

• Privacy & IP: local/secure processing; allows you to purge data and generated text.

Workflow (how you’d actually use it)

1. Load inputs: PDFs, datasets, rubric/guide, style samples.

2. Auto-organize: study matrix + tags + dedupe; you skim the dashboard and approve clusters.

3. Lit synthesis: per theme, read thesis–antithesis–synthesis, approve/edit, and mark key cites.

4. Hypotheses & model: review the proposed RQs/hypotheses and conceptual figure; accept or tweak.

5. Method lock-in: confirm variables, indices, and analysis plan derived from your actual data.

6. Run analysis: execute scripts; inspect QA flags (e.g., missing data, outliers, attention checks).

7. Draft chapters: generate sections with citations, tables/figures, and bridging text.

8. Polish: refine tone, tighten claims, finalize implications; export Word/LaTeX + slide deck.

9. Archive: save the evidence trail (study matrix, scripts, outputs) for viva defense and appendices.

Deliverables

• Annotated outline (chapter by chapter with key sources and “so-what” bullets).

• Full thesis draft (with tracked changes).

• Figures & tables (editable) + captions.

• APA 7 reference list (verified against your PDFs).

• Appendices: study matrix, PRISMA-lite log (if you tracked screening), analysis scripts, robustness notes.

• Defense pack: 15–20-minute slide deck with speaker notes and likely examiner questions.

Customization (example: your Web3 in sport thesis)

• Built-in tags for Ownable (blockchain/NFT/ticketing), Personalised (AI), Immersive (AR/VR).

• Templates for benefit taxonomy, BWS attributes, and a community-health pathway (identification, participation, retention, advocacy).

• Case-sheet generator (Sorare, NBA App personalisation, NFL ALL DAY/Ticketmaster, AR/VR broadcasts) using only the PDFs you provided.

Quality controls

• Evidence density meter: warns if a section leans too much on one source or non-peer-reviewed items.

• Balance & fairness: prompts for counter-evidence to avoid one-sided arguments (antithesis coverage).

• Claim → Data hooks: every empirical claim links to a specific statistic in your dataset or a quoted passage in a PDF.

• Readability & cohesion: checks transitions, paragraph aims, and redundancy; suggests cuts to meet word limits.

Export & interoperability

• Word (.docx) with auto-generated ToC, Lists, headers/footers meeting school formatting.

• LaTeX template with biblatex for APA.

• Citations: BibTeX/CSL JSON for reference managers.

• Data & code: zipped folder with CSVs, scripts, and rendered outputs for reproducibility.