I designed an AI native system that turns messy documents into structured, reviewable work — without losing human control.
Norric is an AI-native operating system that helps private market firms respond to institutional DDQ and RFP requests — by structuring how answers are reused, reviewed, and approved. As a Product Designer & AI Product Strategist, my role was not to “add AI,” but to design a system where AI supports speed and consistency, while humans retain ownership over judgment and risk. This project was about designing boundaries, not full automation for its own sake.
TIMELINE
5 months (2025-2026)
ROLE
Product shaping, IA, interaction, design system, AI assisted review experience
COLLABORATORS
CEO, Front/Back-end developer
COMPANY
Norric Inc.
starting point
When the hardest part isn’t diligence questions, but everything around them.
Investor DDQs aren’t hard because the questions are new. They’re hard because teams answer the same things over and over, across emails, PDFs, and old folders. People spent more time finding past answers than actually reviewing them.
• 60% of DDQ questions repeat
• Prior answers are scattered across PDFs, desktops, SharePoint, and outdated repositories
• IR teams spend hundreds of hours per quarter searching, copying, and rebuilding content
What we did
We didn’t change the questions, we changed how teams respond in their worklife.
We didn’t try to change how investors ask questions. We focused on helping teams respond faster, more consistently, and with less manual work.
The goal was simple: one place to manage DDQs from start to finish.
Norric's main Value
How Norric works
I designed a workflow where each investor request becomes a structured project.
Teams can bring incoming DDQs and related documents into one place and immediately see every question laid out clearly, instead of jumping between files.
They can move through each request step by step — knowing who is responsible for which question, what’s still in progress, and what’s already been reviewed and approved.
As answers take shape, teams can rely on past responses without losing context, while keeping a clear record of decisions and changes along the way.
The result is a process that feels manageable and predictable, even when multiple investor requests are happening at the same time.
How AI agent works
Where AI helps, and where it steps back
AI acts as an assistant inside the workflow, not a replacement for judgment. It proactively finds relevant past answers, suggests drafts when similar questions appear, and provides context through sources and confidence signals.
Norric sits between manual diligence and full automation, using AI to prepare and organize work while keeping decisions firmly in human hands.
Once a response enters review, AI steps out. Final wording, approval, and submission are always human-driven, making accountability explicit rather than hidden behind automation.
THe result
And now with Norric, teams could:
AI acts as an assistant inside the workflow, not a replacement for judgment. It proactively finds relevant past answers, suggests drafts when similar questions appear, and provides context through sources and confidence signals.
Respond to investor requests faster and accurate
Reduce repetitive writing
Stay aligned across IR, legal, and ops
THe Collaboration
Turning ambiguity into shared decisions
I worked directly with the CEO to turn open-ended product goals into concrete workflows the team could actually build against. With engineers, collaboration started early. We defined data structures, edge cases, and system constraints together before designs were finalized, so decisions were made once and carried through implementation.
Instead of handing off finished screens, I used working prototypes to think alongside the team, iterate in real time, and adjust flows as requirements evolved.
reflection
Looking back,
Takeaway 1
Complex workflows don’t need more automation,
they need better structure.
The biggest gains came from clarifying ownership, review states, and information flow, not from adding more features.
Takeaway 2
AI is most effective when its boundaries are explicit.
Using AI to prepare, surface, and suggest work — while keeping decisions human-led — made the system faster without introducing risk.
Takeaway 3
Designing for trust matters more than designing for speed.
In high-stakes workflows, transparency, source traceability, and review history are what make AI usable in practice.