Savvy V2
A decision-support layer that ranks by what locks in if you wait
Savvy is a decision-support layer for wholesale B2B operations. It surfaces the decisions that need intervention before delay becomes loss.
Project type
Self-directed case study
Role
Product Designer, end-to-end
Tools
Figma
Claude
Duration
12 weeks
Summary
Savvy is a decision-support layer for wholesale B2B operations. It surfaces the few decisions that will cost you if you wait, and pushes everything else out of view.
In these operations, losses rarely come from wrong decisions. They come from late ones. The information exists, but it's scattered across too many tools, and nothing signals what needs attention right now. Savvy makes one question, "what needs me today?", the first thing you see instead of something you have to go looking for.
01 THE PROBLEM
Where the losses actually come from
B2B operations don't fail from a lack of information. They fail when decisions happen too late. The information already existed. The problem was catching the moment a decision needed intervention before the delay became expensive. That moment belongs to the deal owner, accountable for revenue and dependent on every team around them.
02 THE REFRAME
The principle I designed around
Most tools answer this with more dashboards and more data. I decided that was the wrong read: this isn't an information problem, it's an attention problem.
So I set one rule for every screen: don't organize the work, reorganize the attention. Each screen had to answer a single question.
03 THE PRODUCT DECISIONS
The three decisions that define Savvy
Everything else on the screen follows from these three calls.
HOW I USED AI
I used AI for the parts that didn't need judgment: testing the scoring logic against edge cases, quickly building rough prototypes so I could see how a layout felt before committing to it, researching how wholesale teams actually track orders, and generating first-pass copy I could react to instead of starting from a blank page.
What I didn't hand off was deciding what the model should weigh, or whether the score should be visible at all. Those are the calls the product lives or dies on, and reviewing an AI-generated decision isn't the same as making it myself.
04 THE PRODUCT
How Savvy turns signals into action
How Savvy turns signals into action
Three screens carry the argument. Each one answers a different question, and each one is the direct result of the calls above.
05 WHERE THIS STANDS
What I'm claiming, and what would prove it
Work platforms already use AI to flag risk and answer prioritization questions when asked. Savvy's difference isn't a feature. It's when it speaks up.
Two numbers would tell me whether that's true.
