Manuscript Builder.
Take a study from "I have results" to a submission-ready draft. IMRaD scaffolds aligned to CONSORT, STROBE, COREQ, PRISMA and ML-conference rubrics, with per-section AI coaching, citation manager, journal-style picker, plain-language summary generator, and a cover-letter drafter.
Templates
Per-template citation style + word target.Title page + Introduction + Methods + Results + Discussion + References. Add the rest as needed.
Standard IMRaD with plain-language summary, ethics, trial registration slot, CRediT contributions.
Tighter IMRaD with explicit CONSORT participant flow, trial registration, expanded ethics.
Abstract → Introduction → Related work → Method → Experiments → Discussion → Conclusion. IEEE-style citations.
Theory-driven structure: theoretical framework before methods, expanded discussion.
Methods structured around CONSORT 2010 items: design, randomisation, blinding, intention-to-treat, harms.
Reflexivity statement, sampling-to-saturation, member checking, thick description.
PRISMA 2020 structure: protocol, search, screening (PRISMA flow), risk of bias, synthesis.
Reporting-guideline rubrics
Each rubric carries its own red-flag list. The AI coach reads them on every action.
CONSORT-aligned reports must specify design (parallel/cluster/crossover), randomisation method, allocation concealment, blinding, primary outcome (single, prespecified), intention-to-treat analysis, attrition, and adverse events. The CONSORT flow diagram is expected as Figure 1.
STROBE reports should name the design in the title, describe the setting and dates, list eligibility criteria precisely, define exposures and outcomes with measurement details, address bias (selection, information, confounding), justify sample size, describe statistical methods including how confounding was addressed, report participants at each stage with a flow diagram, and discuss limitations openly.
COREQ has 32 items across three domains: research team & reflexivity (who interviewed, their training, prior relationship with participants), study design (theoretical framework, sampling, sample size, saturation), and analysis & findings (data analysis approach, coding, themes derived, quotation usage, member checking).
PRISMA 2020 requires explicit protocol registration (PROSPERO), comprehensive search strategy with date + databases, eligibility criteria (PICOS), PRISMA flow diagram, risk-of-bias assessment with named tool (RoB 2, ROBINS-I, Newcastle-Ottawa), and a clear synthesis description.
ML papers reward a clear contribution stated in the second sentence, a related-work section that contrasts (not just cites) prior work, reproducibility (code + data + hyperparameters), baselines with statistical significance, an ethics / broader-impact statement, and a candid limitations section.
Social-science papers expect a clearly stated puzzle, engagement with named theoretical schools, transparent operationalisation of constructs, reflexive discussion of researcher positionality where applicable, and an explicit statement of theoretical contribution alongside empirical findings.
Apply standard IMRaD conventions: each section leads with its strongest sentence, claims are calibrated to the evidence, methods are reproducible from the description, results are reported with effect sizes + uncertainty, discussion engages prior literature, and limitations are stated honestly.
The AI rewrites and tightens using only what you supply. It never invents sample sizes, p-values, citations or co-authors — those become [bracketed placeholders] for you to fill in.
Wrote the abstract in Abstract Studio? Paste it in. Defended the methods in Methods Memo? Use the same prose. Got reviews back? Send the response with Reviewer Response Studio.