Designs
Empty state
Select files to upload. Disable files that aren't supported.
Documents uploaded
Bulk select files to AI generate work instructions
Processing in the background
Processing files
Generated Drafts - Click draft to view document
Learnings & Results
For video -> work instruction
The loud environment of these manufacturing facilities caused errors in the AI generated documents. We were using AWS transcribe to get text from the video. AWS had trouble transcribing text when there was a lot of background noise.
For PDF, DOCX, PPTX -> work instruction
There were even more variables with non-video files. Trying to break down regular documents into Dozuki format caused a lot of issues in the beginning, so the team decided to ship video to work instruction first to test our proof of concept.
Introduction
One of our big goals for the product team was to reduce the time for enterprise customers to get onboarded. The current process took over 9 months.
Problems
Customers had 1000s of existing paper documents and legacy media files that they operate and train employees from. Turning this into Dozuki documents took over 9 months to get them fully created.
Manual Creation of Documents - Customer support team had a division to turn customer's PDFs, word docs, and other documents into Dozuki documents.
Opportunities
User opportunities (Why do users care?)
Reduce the time creating converting documents into Dozuki for efficiency and productivity.
Business Opportunities (What's in it for Dozuki?)
Faster onboarding time meant customers start getting value out of Dozuki and reduce the time to upsell customers.
Document Conversion Flows
Upload file -> Generate Doc
AI Document Creation
Turning videos and other freeform files into Dozuki formatted work instructions
Impact
Reduced onboarding time by 3 months
+$500k ARR
Role / Team
Product Design
0 -> 1 feature
Me + PM + 4 Engineers
Timeline
2 weeks research
1 week of design
Full case study available during portfolio presentation