A focused path from systematic-review screening to a paid public beta, built one phase at a time. Ship the differentiator, then expand.
The MVP must let an early adopter complete a real review and cite Orivion in the methods section. Everything else is expansion.
Honest math at 5–10 hrs/week. Closed beta first to learn, public beta when payments and the differentiating feature ship.
Sequenced for revenue first, then expansion. The MVP ends at Phase 5. Phases 6–10 are the year after.
Blind mode → Conflict resolution → Realtime polish. Don't skip ahead.
Complete the title/abstract story before touching anything else. Plan first (algorithmic + ML + AI hybrid), then build.
Small effort, huge value. Required for SR publication. Lets early users finish a real review and cite Orivion — even if full-text is done elsewhere temporarily.
The gate for Paddle / Polar approval. Do it once, properly.
Stripe checkout + webhooks → subscription state + feature gating → submit Paddle / Polar. Can't validate willingness-to-pay without this.
PDF support (upload + view) → Full-text finder → Full-text screening UI. Annotation overlay is enough — PDF editing not worth it for v1.
Structured extraction forms tied to study designs. Outcome variables, effect sizes, study characteristics.
Cochrane RoB 2, ROBINS-I, and other tools as structured assessments per included study.
Big architectural decision. Likely Supabase Edge Function or containerized R service. Evaluate WebR (R compiled to WASM) — runs client-side, no infra cost, no API tokens.
Biggest, riskiest build. Defer until paying users tell you they want this in-app rather than in Word / Google Docs.
The original task list had features. These are the things that decide whether the launch actually works.
Non-negotiable for beta
Worth considering
Don't fly blind
Lock the MVP list. Scope creep is the biggest risk — every "while I'm in there" addition costs a week.
Automated screening is non-negotiable. It is the differentiator that separates Orivion from a free Rayyan clone.
Closed beta first, public beta second. Use the gap to fix what real users break.
A delayed launch with the differentiator beats an on-time launch without it.