Project intro
SpotDraft's customers use metadata extraction using AI for generating reports and for glancing through contracts for a summary. To get to these metadata they have to setup fields and a prompt for AI to use for each field in a separate setup process. Most of this setup is currently done by SpotDraft's legal tech teams. Even for increasing extraction accuracy legal tech team jumps in to edit the prompt in custom ways for each customer.
This project shows how the product, legal tech team and design team built a self serve way for customers to setup and iterate their fields-prompts.
Contribution
My contribution included discovery, design handoff and quality assurance.
User journeys
8 separate steps to set up the integration.
Too much emphasis on text and explanations over actionables
Not native friendly : Tray’s interface feels separate from SpotDraft, causing a fragmented user experience. Users had to map fields that were not mandatory to be mapped since Tray provides a more generic integration setup that a lot of other customers also use.
Steep Learning Curve and Overwhelming Navigation: Setting up the integration required navigating complex workflows, resulting in setup times exceeding 120 mins and frustrating non-technical users.
Scalability Issues: Each new integration demands building separate workflows, making the system fragile and difficult to scale as business needs evolve, increasing maintenance overhead.
Like what you see?
Resume
Contact
+91-9521788932
arpitsingh10c@gmail.com
Made on Framer & Figma
in Bengaluru, India
By Arpit Singh
Making AI Extraction Fully Self-Serve
Empowering in house legal teams with help of a prompt library and an AI playground to setup and improve AI powered contract metadata extraction in SpotDraft.
6 months
