Requirements
Video transcript
00:00 — Writing requirements with ReqGen
Writing requirements with ReqGen. From a blank page to a reviewed specification.
00:09 — Two ways to start
Generate with AI drafts the complete document from one prompt — every applicable V-model category in a single pass. Build it yourself starts from a blank document or a CSV import, with the AI agent on call to interview you and help draft, category by category.
00:33 — Document Info
Set author, company, status, and the document's purpose & scope up front — both paths share it, and exports carry it. The agent can record purpose/scope for you after its interview.
00:50 — Your AI agent
The agent pins to whatever document you have open. Ask it to interview you, draft requirements, or rename the document — it batches agreed changes into a follow-on that you review in the diff panel and merge. Attach specs and it grounds its drafts in them.
01:12 — Review and refine
Your requirements appear here, grouped by category. Edit inline, add via the category cards, or flip Follow-on Mode on to run a one-shot AI follow-on — then review the changes and merge. Baselines, revisions, and reviews all live on this page.
01:35 — docs.creopus.ai
Full documentation, including every tool, is at docs.creopus.ai
The first artifact on every node — everything else derives from it.

What it's for
ReqGen drafts a structured requirement set scoped to the hierarchy node you are on. The main view is a grid of category cards; “View All” opens a full table with a live-count tab per category (Core, Constraints/Risks/Assumptions, Non-Functional, Interfaces, Hardware, Testing, Traceability, plus Medical / Automotive when those domain packs are on). Every requirement gets a self-describing ID and can carry compliance columns (DAL / ASIL / SIL / SW Class).
When to use it
First on every node. Do this before deriving any downstream artifact.
How it works
- Select a System Hierarchy node, then pick a path — “Generate with AI” or “Build it yourself”. → Ancestor requirements load as context so output is derived, not invented; the manual path offers a blank doc or a CSV import with the agent on call.
- Open the category / domain picker and toggle domain packs. → Medical adds ISO 13485 / 21 CFR 820 categories, Automotive adds ASPICE / ISO 26262 ASIL; at system/root level a short clarify questionnaire (ARP4754A stakeholder needs) runs before generation.
- Describe the system and click Generate Requirements. → Category cards fill with auto-IDed requirements (BUS-001, FUNC-012…), each with a priority dot; “View All” opens the tabbed full table.
- Open a requirement (its ID is a link) to edit it, or use the per-row discussion. → Inline editors per field; soft-delete strikes a row through with Undo, and a permanent purge appears on the main doc once a baseline exists. AI docs are read-only unless you turn on Follow-on mode.
- Open the Parent Requirements panel and link across levels. → Browse ancestor nodes’ requirement docs, multi-select, and “Link to Selected” — creating cross-level traceability links that surface in the Traceability tab.
- Search, then Export. → Export to PDF, stamped PDF, DOC, CSV, TXT or JSON; a ReqGen CSV round-trips back in via Import.
Mechanics
- Category-card grid + tabbed “View All” full table (V-model taxonomy)
- Domain packs (Medical, Automotive) + optional compliance columns (DAL/ASIL/SIL/SW Class)
- Self-describing IDs · per-field inline edit · soft-delete → Undo → purge
- Traceability Matrix + cross-level Parent-Requirements linking
- Per-item discussion · per-item review Sign-off · pre-gen clarify (root level)
- Export: PDF · stamped PDF · DOC · CSV · TXT · JSON