Turn prompt-engineering research into audit-ready evidence.

R&D incentive regimes — following the OECD's Frascati Manual framing — ask for the same things everywhere: evidence of a technological uncertainty that a competent professional could not resolve from existing knowledge, a systematic process of experimentation to resolve it, and records made at the time, not reconstructed at filing season. Systematic prompt-engineering research fits this frame naturally — but only if the paper trail exists.

AIScholar's R&D module builds that trail as a by-product of the pipeline you already use: every design created, run launched, coding pass completed, and analysis executed is logged automatically, with timestamps, while you work.

• Uncertainty statements

  Versioned statements of the technological uncertainty your work addresses, with AI-assisted drafting and an AI quality evaluation that critiques weak or generic claims before an auditor does.

• Contemporaneous activity log

  Auto-generated entries for every pipeline action, plus manual entries with notes for work done outside the platform. Exportable as JSON and CSV.

• Effort tracking

  A floating start/stop timer on every pipeline page, manual time entry, weekly summaries by activity category — investigation & design, experimentation, analysis, support — and CSV timesheet export.

• Knowledge progress tracker

  Records what was learned against each uncertainty statement, with evidence links, remaining uncertainty, and resolution status — the "systematic progress" auditors look for, with AI-drafted summaries.

• Inactivity nudges

  A seven-day inactivity threshold triggers a gentle banner so documentation gaps are caught in the week they happen, not the year after.

• ZIP evidence package

  One export bundles the summary markdown, uncertainty statements, activity log (JSON + CSV), knowledge progress, and an effort & expenditure summary with a categorization guide.

1. 1

   Enable R&D mode per project

   Flip the toggle on any project where the work may qualify. A clear legal disclaimer sets expectations, and R&D endpoints are enforced server-side only while the mode is on.

2. 2

   State the uncertainty early

   Draft the technological uncertainty statement at project start — with AI assistance — so the record shows the uncertainty preceded the experiments that resolved it.

3. 3

   Work; the log writes itself

   Run your study as usual. The activity log accumulates automatically; the timer captures effort; the knowledge tracker records findings as they land.

4. 4

   Export for your advisor

   At reporting time, download the ZIP package and hand it to your tax advisor or grant office — organized, timestamped, and categorized.

The R&D trail pairs naturally with the scientific audit trail in reproducibility — one records how the research was done, the other why it counts as research.