AI Spec Generator Chat
Author and refine business process specifications in a threaded AI chat with open questions, reference-file context, and one-click finalization.
Each draft business process in OL Spec has a dedicated spec generator chat (/business-process/<id>/chat) where teams describe requirements, iterate with an AI assistant, and finalize a rich specification.
Capabilities
Composer
- Free-form messages; each Send appends to the thread
- Optional reference files on the next message only — PDF, Word (
.docx), or images (cleared after a successful send) - Limits: up to 12 files, 20 MB each, 48 MB total per send
- Send Final — appends a finalization instruction so the model self-suggests defaults and must not include an Open questions section
- New chat — clears the thread for this business process
Open questions panel
When the assistant’s last reply contains an Open questions heading with a numbered or bulleted list, items are extracted above the composer:
- Copy all — clipboard
- Append to message — paste into the composer
- Suggest — hidden instruction asking the model to propose its own answers and return a finalized spec
Conversation
- Full thread on the right; latest assistant reply is auto-saved as the specification content
- Open business process appears once a save has happened
- Backend guide dialog explains how the backend interprets the spec
- Chat direction follows the workspace language (including RTL)
Typical workflow
- Create a new business process (name + optional description) → draft is created and chat opens