How it works
What this workflow does
Turn a spoken project debrief into a Webflow case-study draft. The editor opens a private n8n form, dictates the client challenge and what changed, enters verified results, and confirms that the material is approved. The workflow validates the input, drafts a title, summary, challenge, approach, and verified outcome with OpenAI, and creates a staged Webflow CMS item.
The workflow imports inactive with dryRun: true. Run with sample data uses a fictional client label and project review, so the preview does not call OpenAI, Webflow, or the Data Table.
This is the case-study layer for the Content Publishing, Social Distribution & Newsletter Intelligence blueprint; the Project Delivery QA & Completion blueprint covers the project closeout that produces the debrief.
Approval before drafting
Receive event is an n8n Form Trigger protected by Basic Auth. The form asks for a Draft ID, an Approved client name or anonymous label, an approval confirmation, Challenge notes, Implementation notes, and Verified results. The approval value must be Yes; anything else stops before the model call.
The draft ID must be a stable lowercase value. A live run also requires a Data Table, model, 24-character Webflow collection ID, and valid distinct CMS field slugs.
Each section keeps its own evidence
The model returns a challenge, approach, and outcome. Each section must contain its own evidence quote: the challenge quote from the challenge notes, the approach quote from the implementation notes, and the outcome quote from the verified results. Quotes and section text are length-checked, and the workflow escapes the content before building the CMS body.
Any number in the copy must also appear in the verified results. The workflow adds the approved client label as the first paragraph and never changes it into a different identity. It creates a staged Webflow item with isDraft: true and verifies the returned item ID, draft state, archive state, and slug before recording draft_ready. It never publishes.
Replays and validation scope
The Data Table is keyed by the draft ID. An existing request stops the run and preserves the current review state. A new request is reserved as draft_pending before the Webflow call.
The supplied report confirms the isolated n8n 2.47.4 import, inactive state, the built-in sample, and a local copy of the native form. It also covers client-approval enforcement, section-specific evidence, unverified number rejection, and staged-only Webflow behavior. Live Wispr Flow dictation, deployed forms, credentials, model responses, and Webflow or Sheets operations require staging.
How to use
Before you start
- A current n8n instance with Form Trigger and Data Table support.
- A Webflow account, site, and collection for case-study drafts.
- A Webflow API token with
cms:write. - An OpenAI API key and a JSON-mode model.
- Wispr Flow configured for the editors who will dictate.
- Written permission for the client material, or an approved anonymous label.
1. Import and run the sample
Download the complete package, extract the ZIP, and read the full setup guide. Import voice-to-webflow-case-study.n8n.json and keep it inactive.
Open Run with sample data and choose Test workflow. Assemble case study draft should return a title, summary, and the three sourced sections with isDraft enabled.
2. Prepare the private form
Select a Basic Auth credential on Receive event. Start the form test and open the Test URL. Dictate the challenge and implementation notes with Wispr Flow, review the inserted text, and submit.
3. Create the state table
Create an n8n Data Table with text columns event_key, status, detail, and result_id. Enter its ID in Setup together with the Webflow collection ID, model, and brand voice.
Set bodyField to the Rich Text field slug and summaryField to the Plain Text summary slug, or clear the summary field. Configure any other required collection fields before staging.
4. Connect OpenAI and Webflow
Add an OpenAI Header Auth credential on Draft with OpenAI. Select a Webflow Header Auth credential on Create Webflow CMS draft with cms:write. The workflow uses the staged-item endpoint; review and publish through your normal editorial process.
5. Test and operate
Provide an approved client name or anonymous label and confirm permission before submitting. Only use information approved for processing by your model provider, including when the case study will be anonymous.
The sample uses a fictional client label and an outcome scoped to a fictional project review. Approval must be Yes. The challenge, approach, and outcome each require an excerpt from their corresponding input, and numerical claims must appear in the verified results. Check the denominator, attribution, and client permissions before publication.
If draft_pending remains, look up uptune-draft-<draft-id> in Webflow before retrying. Store the item ID and mark the record ready if the draft exists. Serialize requests for the same ID and use a new ID for a revision.
Related automations
Use the Voice-to-Webflow Article Draft System for founder notes and the Voice-to-Webflow Release Notes System for product updates. The Client Onboarding & Project Handoff Automation supplies the handoff context behind many case studies. Read the Content Publishing & Newsletter Intelligence blueprint for the wider content process.